CN117744033A - Data fluctuation early warning method and device, nonvolatile storage medium and electronic equipment - Google Patents
Data fluctuation early warning method and device, nonvolatile storage medium and electronic equipment Download PDFInfo
- Publication number
- CN117744033A CN117744033A CN202311801387.6A CN202311801387A CN117744033A CN 117744033 A CN117744033 A CN 117744033A CN 202311801387 A CN202311801387 A CN 202311801387A CN 117744033 A CN117744033 A CN 117744033A
- Authority
- CN
- China
- Prior art keywords
- data
- sample
- data set
- preset
- time period
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 105
- 238000003860 storage Methods 0.000 title claims abstract description 31
- 238000004458 analytical method Methods 0.000 claims abstract description 51
- 238000012544 monitoring process Methods 0.000 claims abstract description 31
- 238000012937 correction Methods 0.000 claims abstract description 10
- 238000012549 training Methods 0.000 claims description 53
- 238000009826 distribution Methods 0.000 claims description 21
- 230000000007 visual effect Effects 0.000 claims description 15
- YHXISWVBGDMDLQ-UHFFFAOYSA-N moclobemide Chemical compound C1=CC(Cl)=CC=C1C(=O)NCCN1CCOCC1 YHXISWVBGDMDLQ-UHFFFAOYSA-N 0.000 description 24
- 230000008569 process Effects 0.000 description 22
- 238000005516 engineering process Methods 0.000 description 16
- 238000012545 processing Methods 0.000 description 15
- 238000010586 diagram Methods 0.000 description 12
- 238000013473 artificial intelligence Methods 0.000 description 11
- 230000000694 effects Effects 0.000 description 9
- 230000002159 abnormal effect Effects 0.000 description 8
- 238000007781 pre-processing Methods 0.000 description 8
- 238000004140 cleaning Methods 0.000 description 7
- 238000007476 Maximum Likelihood Methods 0.000 description 6
- 238000013480 data collection Methods 0.000 description 6
- 238000013485 heteroscedasticity test Methods 0.000 description 6
- 238000013135 deep learning Methods 0.000 description 5
- 238000010801 machine learning Methods 0.000 description 5
- 230000007246 mechanism Effects 0.000 description 5
- 238000007619 statistical method Methods 0.000 description 5
- 238000012800 visualization Methods 0.000 description 5
- 238000005311 autocorrelation function Methods 0.000 description 4
- 230000008859 change Effects 0.000 description 4
- 230000007613 environmental effect Effects 0.000 description 4
- 238000009499 grossing Methods 0.000 description 4
- 238000005065 mining Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 238000007405 data analysis Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 230000006978 adaptation Effects 0.000 description 2
- 230000006399 behavior Effects 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000010168 coupling process Methods 0.000 description 2
- 238000005859 coupling reaction Methods 0.000 description 2
- 230000001419 dependent effect Effects 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000009977 dual effect Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 230000036962 time dependent Effects 0.000 description 2
- 230000005856 abnormality Effects 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000013450 outlier detection Methods 0.000 description 1
- 238000011946 reduction process Methods 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
- 238000012731 temporal analysis Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000000700 time series analysis Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a data fluctuation early warning method and device, a nonvolatile storage medium and electronic equipment. Wherein the method comprises the following steps: acquiring a historical data set of a historical time period; analyzing the historical data set by using a preset prediction model to obtain a first prediction data set of a target time period, wherein the preset prediction model at least comprises: a preset autoregressive moving average model for predicting a second prediction data set of the target time period and a preset autoregressive conditional heteroscedastic model for predicting an error data set of the target time period, the first prediction data set being a correction result of the second prediction data set based on the error data set; monitoring the degree of data fluctuation of the predicted dataset relative to the historical dataset; and carrying out fluctuation early warning under the condition that the fluctuation degree of the data exceeds a fluctuation threshold value. The invention solves the technical problem that the prior art cannot carry out fluctuation analysis and early warning on data aiming at a large-scale data environment.
Description
Technical Field
The invention relates to the field of big data processing, in particular to a data fluctuation early warning method and device, a nonvolatile storage medium and electronic equipment.
Background
The fluctuation analysis early warning is a method for monitoring and predicting fluctuation conditions in the fields of markets, economy, environments and the like and sending early warning timely. The existing fluctuation analysis early warning method is mostly based on the traditional statistical technology and experience judgment, and is difficult to deal with large-scale and dynamic data environments.
Aiming at the problem that the prior art cannot carry out fluctuation analysis and early warning on data aiming at a large-scale data environment, no effective solution is proposed at present.
Disclosure of Invention
The embodiment of the invention provides a data fluctuation early warning method and device, a nonvolatile storage medium and electronic equipment, which at least solve the technical problem that the prior art cannot perform fluctuation analysis early warning on data in a large-scale data environment.
According to an aspect of the embodiment of the present invention, there is provided a method for early warning of fluctuation of data, including: a historical dataset of a historical time period is obtained, wherein the historical time period comprises: a plurality of historical moments, the historical dataset comprising: historical data corresponding to a plurality of historical moments; analyzing the historical data set by using a preset prediction model to obtain a first prediction data set of a target time period, wherein the preset prediction model at least comprises: the method comprises the steps of presetting an autoregressive moving average model and a preset autoregressive conditional covariance model, wherein the preset autoregressive moving average model is used for predicting a second prediction data set of the target time period according to the historical data set, the preset autoregressive conditional covariance model is used for predicting an error data set of the target time period, and the target time period comprises: a plurality of target moments, the first prediction dataset comprising: at least one first prediction data corresponding to the target moment, wherein the second prediction data set comprises: at least one second prediction data corresponding to the target moment, wherein the error data set comprises: at least one error data corresponding to the target time, wherein the first prediction data is a correction result of the second prediction data and the error data of the same target time; monitoring a degree of data fluctuation of the predictive dataset relative to the historical dataset; and carrying out fluctuation early warning under the condition that the fluctuation degree of the data exceeds a fluctuation threshold value.
Optionally, before analyzing the historical data set by using a preset prediction model to obtain a first prediction data set of the target time period, the method further includes: obtaining a preset sample data cluster of a preset sample time period, wherein the preset sample time period at least comprises a first sample time period, a second sample time period and a third sample time period which are continuous, and the first sample time period comprises: a plurality of first sample times, the second sample time period comprising: a plurality of second sample times, the third sample time period comprising: a plurality of third sample moments, the preset sample data clusters at least comprise: a first sample data set for the first sample time period, a second sample data set for the second sample time period, and a third sample data set for the third sample time period, the first sample data set comprising: a plurality of first sample data corresponding to the first sample time, and the second sample data set includes: and a plurality of second sample data corresponding to the second sample moments, wherein the third sample data set comprises: third sample data corresponding to a plurality of third sample moments; training the preset autoregressive moving average model using a first sample data cluster, wherein the first sample data cluster comprises: the first sample data set and the second sample data set; analyzing the first sample data cluster by using the preset autoregressive moving average model to obtain a second sample data cluster, wherein the second sample data cluster comprises: a fourth sample data set for the second sample time period, the fourth sample data set being a prediction result based on the first sample data set, and a fifth sample data set for the third sample time period, the fifth sample data set being a prediction result based on the second sample data set, the fourth sample data set comprising: fourth sample data corresponding to the second sample moments, wherein the fifth sample data set comprises: fifth sample data corresponding to the third sample moments; training the preset autoregressive conditional heteroscedastic model using a preset sample residual data cluster, wherein the preset sample residual data cluster comprises: a first set of residual data for the second sample time period and a second set of residual data for the third sample time period, the first set of residual data comprising: first residual data corresponding to a plurality of second sample moments, wherein the first residual data is a difference value between the fourth sample data and the second sample data at the same second sample moment, and the second residual data set comprises: and second residual data corresponding to the third sample time, wherein the second residual data is a difference value between the fifth sample data and the third sample data at the same third sample time.
Optionally, using the first sample data cluster, training the preset autoregressive moving average model includes: analyzing the first sample data set using an initial autoregressive moving average model to obtain a sixth sample data set of the second sample time period, wherein the sixth sample data set comprises: a plurality of sixth sample data corresponding to the second sample time; comparing the difference between the sixth sample data set and the second sample data set to obtain a first comparison result; and adjusting a first model parameter set in the initial autoregressive moving average model according to the first comparison result to obtain the preset autoregressive moving average model.
Optionally, training the preset autoregressive conditional heteroscedastic model using a preset sample residual data cluster comprises: determining residual errors of the second sample data cluster and the first sample data cluster to obtain a preset sample residual error data cluster; detecting whether preset residual data in the preset sample residual data cluster accords with white noise distribution, wherein the preset residual data at least comprises: the first residual data and the second residual data; and under the condition that the preset residual data accords with the white noise distribution, training the preset autoregressive condition heteroscedastic model by using the preset sample residual data set.
Optionally, training the preset autoregressive conditional heteroscedastic model using a preset sample residual data cluster comprises: analyzing the first residual data set by using an initial autoregressive conditional heteroscedastic model to obtain a third residual data set of the third sample time period, wherein the third residual data set comprises: third residual data corresponding to a plurality of third sample moments; comparing the difference between the third residual data set and the second residual data set to obtain a second comparison result; and adjusting a second model parameter set in the initial autoregressive condition heteroscedastic model according to the second comparison result to obtain the preset autoregressive condition heteroscedastic model.
Optionally, monitoring the extent of data fluctuation of the predictive data set relative to the historical data set comprises: determining residuals of the prediction data set and the historical data set to obtain a prediction residual sequence, wherein the prediction residual sequence comprises: the prediction residual data are arranged according to a time sequence, and each prediction residual data is the residual of the predicted first prediction data and the preset historical data at the target moment; determining a statistical index of the prediction residual sequence, wherein the statistical index at least comprises: standard deviation, variance and mean; and evaluating the fluctuation degree of the data according to the statistical index.
Optionally, after monitoring the degree of data fluctuation in the predicted data set and the actual data set for the target time period, the method further comprises: displaying the degree of fluctuation of the data by using a visual chart, wherein the visual chart at least comprises: graph, report, dashboard, thermodynamic and dynamic charts.
According to another aspect of the embodiment of the present invention, there is also provided a device for early warning of fluctuation of data, including: the system comprises an acquisition module, a storage module and a storage module, wherein the acquisition module is used for acquiring a historical data set of a historical time period, and the historical time period comprises: a plurality of historical moments, the historical dataset comprising: historical data corresponding to a plurality of historical moments; the analysis module is configured to analyze the historical data set by using a preset prediction model to obtain a first prediction data set of a target time period, where the preset prediction model at least includes: the method comprises the steps of presetting an autoregressive moving average model and a preset autoregressive conditional covariance model, wherein the preset autoregressive moving average model is used for predicting a second prediction data set of the target time period according to the historical data set, the preset autoregressive conditional covariance model is used for predicting an error data set of the target time period, and the target time period comprises: a plurality of target moments, the first prediction dataset comprising: at least one first prediction data corresponding to the target moment, wherein the second prediction data set comprises: at least one second prediction data corresponding to the target moment, wherein the error data set comprises: at least one error data corresponding to the target time, wherein the first prediction data is a correction result of the second prediction data and the error data of the same target time; a monitoring module for monitoring the degree of data fluctuation of the predicted data set relative to the historical data set; and the early warning module is used for carrying out fluctuation early warning under the condition that the fluctuation degree of the data exceeds a fluctuation threshold value.
According to another aspect of the embodiment of the present invention, there is further provided a nonvolatile storage medium, where the nonvolatile storage medium is used to store a program, and when the program runs, a device where the nonvolatile storage medium is controlled to execute the method for early warning fluctuation of the data.
According to another aspect of the embodiment of the present invention, there is also provided an electronic device, including: the system comprises a memory and a processor, wherein the processor is used for running a program stored in the processor, and the program executes the fluctuation early warning method of the data when running.
In the embodiment of the invention, the preset prediction model can be determined by utilizing an artificial intelligence technology through algorithms such as machine learning, deep learning and the like and is used for mining potential rules and trends from data, and the preset prediction model combining the preset autoregressive moving average model and the preset autoregressive conditional heteroscedure model has the capability of processing high-frequency data, considering the problem of heteroscedasticity and the like, the second preset data set of a target time period can be predicted according to a historical data set through the preset autoregressive moving average model, the error uncertainty of a prediction result can be considered through the preset autoregressive conditional heteroscedure model, the second preset data set can be corrected, a more accurate second prediction data set can be obtained, further, the data fluctuation degree of the target time period can be predicted based on the second prediction data set, and early warning can be carried out according to the predicted data fluctuation degree, so that the technical effects of intelligent prediction and early warning of fluctuation conditions are realized, and the technical problem that the prior art cannot carry out data fluctuation analysis on a large-scale data environment is solved.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this application, illustrate embodiments of the invention and together with the description serve to explain the invention and do not constitute a limitation on the invention. In the drawings:
FIG. 1 is a flow chart of a method for early warning of fluctuations in data according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a surge analysis early warning method according to an embodiment of the present invention;
FIG. 3 is a schematic illustration of an ARIMA-GARCH model training process in accordance with an embodiment of the present invention;
FIG. 4 is a schematic illustration of training an ARIMA model in accordance with an embodiment of the present invention;
FIG. 5 is a schematic diagram of a data fluctuation pre-warning device according to an embodiment of the present invention;
fig. 6 is a block diagram of a computer terminal according to an embodiment of the present invention.
Detailed Description
In order that those skilled in the art will better understand the present invention, a technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in which it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention.
It should be noted that the terms "first," "second," and the like in the description and the claims of the present invention and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the invention described herein may be implemented in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
According to an embodiment of the present invention, there is provided an embodiment of a method for early warning of fluctuations in data, it being noted that the steps shown in the flowcharts of the drawings may be performed in a computer system such as a set of computer executable instructions, and that although a logical order is shown in the flowcharts, in some cases the steps shown or described may be performed in an order different from that herein.
Fig. 1 is a flowchart of a method for early warning of fluctuation of data according to an embodiment of the present invention, as shown in fig. 1, the method includes the steps of:
step S102, acquiring a historical data set of a historical time period, wherein the historical time period comprises: the plurality of historical moments, the historical dataset comprising: historical data corresponding to a plurality of historical moments;
step S104, analyzing the historical data set by using a preset prediction model to obtain a first prediction data set of a target time period, wherein the preset prediction model at least comprises: the method comprises the steps of presetting an autoregressive moving average model and a preset autoregressive condition heteroscedastic model, wherein the preset autoregressive moving average model is used for predicting a second prediction data set of a target time period according to a historical data set, the preset autoregressive condition heteroscedastic model is used for predicting an error data set of the target time period, and the target time period comprises: the plurality of target moments, the first prediction dataset comprising: the first predicted data corresponding to the at least one target time, the second predicted data set comprising: the second prediction data corresponding to the at least one target moment, and the error data set comprises: at least one error data corresponding to the target moment, wherein the first predicted data is the correction result of the second predicted data and the error data of the same target moment;
Step S106, monitoring the data fluctuation degree of the predicted data set relative to the historical data set;
and step S108, carrying out fluctuation early warning under the condition that the fluctuation degree of the data exceeds a fluctuation threshold value.
In the embodiment of the invention, the preset prediction model can be determined by utilizing an artificial intelligence technology through algorithms such as machine learning, deep learning and the like and is used for mining potential rules and trends from data, and the preset prediction model combining the preset autoregressive moving average model and the preset autoregressive conditional heteroscedure model has the capability of processing high-frequency data, considering the problem of heteroscedasticity and the like, the second preset data set of a target time period can be predicted according to a historical data set through the preset autoregressive moving average model, the error uncertainty of a prediction result can be considered through the preset autoregressive conditional heteroscedure model, the second preset data set can be corrected, a more accurate second prediction data set can be obtained, further, the data fluctuation degree of the target time period can be predicted based on the second prediction data set, and early warning can be carried out according to the predicted data fluctuation degree, so that the technical effects of intelligent prediction and early warning of fluctuation conditions are realized, and the technical problem that the prior art cannot carry out data fluctuation analysis on a large-scale data environment is solved.
In the step S102, the historical data set may be collected, processed and analyzed in real time based on the big data technology, and more accurate and comprehensive fluctuation information may be provided based on the massive historical data set.
In the step S102, in the process of acquiring the historical data set, preprocessing such as data cleaning, outlier detection, data smoothing and the like can be adopted for the collected data, so that the quality and the integrity of the data are ensured, and a reliable data base is provided for subsequent fluctuation analysis and early warning.
It should be noted that, in the data in the preset scene, the collected specific data can ensure accuracy, and the specific data does not need to be preprocessed, where the specific data includes: data collected by the real-time sensor, data derived from the database, and data provided by the high reliability data source.
In the above step S104, the autoregressive moving average model ARIMA is: ARIMA (p, d, q),wherein p is an autoregressive term, d is a differential number and q is a moving average term, which can be determined empirically; b is denoted as the fallback operator: b (B) k X t ＝X t-k The method comprises the steps of carrying out a first treatment on the surface of the Phi is a parameter in an autoregressive model AR, and represents the influence degree of historical data of an ith time period on current data in the autoregressive model, and is usually estimated by a statistical method such as a least square method; θi is a parameter in the moving average model MA, and represents the degree of influence of the error of the ith time period on the current data in the moving average model, and is usually estimated by a statistical method such as a least square method.
In the above step S104, the autoregressive conditional heteroscedastic model GARCH is: GARCH (P, Q), σt 2 ＝ω+∑αiεt-i 2 +∑βiσt-i 2 Wherein P is an autocorrelation term for measuring ε t-i 2 The number of epochs (i.e., past volatility information); q is a moving average term used to measure σt-i 2 The number of epochs (i.e., the volatility information of past prediction errors); omega is a constant term; εt is the residual, i.e., the difference between the actual data and the predicted data; sigma t 2 Is the fluctuation rate or variance of time t; αi is shown in GARIn the CH model, the influence degree of the residual square of the ith time period on the current fluctuation rate is estimated based on a maximum likelihood estimation method; βi represents the degree of influence of the fluctuation rate of the ith time period on the current fluctuation rate in the GARCH model, and is also estimated by the maximum likelihood estimation method.
In the step S104, the prediction model is preset to combine the constancy of the ARIMA model and the heteroscedasticity of the GARCH model to form a more powerful model.
In the step S104, the degree of data fluctuation may be determined according to the residual error between the predicted data set and the preset historical data in the historical data set, where the preset historical data may be set based on the historical data set, such as average data, mode data, median data, and the like of the historical data set; the preset history data may be set according to a preset history data set in a target history period adjacent to the target time period.
The history data, the prediction data, the error data, and the residual data may be data values.
In the step S108, the wave motion warning includes: an alert or notification is sent to the relevant person to take action in time, for example, the alert, message, email or other notification means may be sent to the relevant person to deliver the pre-warning information.
As an alternative embodiment, before analyzing the historical data set using the preset prediction model to obtain the first predicted data set of the target time period, the method further includes: obtaining a preset sample data cluster of a preset sample time period, wherein the preset sample time period at least comprises a first sample time period, a second sample time period and a third sample time period which are continuous, and the first sample time period comprises: the plurality of first sample moments, the second sample time period comprising: the plurality of second sample times, the third sample time period comprising: the preset sample data clusters at least comprise a plurality of third sample moments: a first sample data set for a first sample time period, a second sample data set for a second sample time period, and a third sample data set for a third sample time period, the first sample data set comprising: the first sample data corresponding to the plurality of first sample moments, the second sample data set including: the second sample data corresponding to the plurality of second sample moments, and the third sample data set includes: third sample data corresponding to a plurality of third sample moments; training a preset autoregressive moving average model using a first sample data cluster, wherein the first sample data cluster comprises: a first sample data set and a second sample data set; analyzing the first sample data cluster by using a preset autoregressive moving average model to obtain a second sample data cluster, wherein the second sample data cluster comprises: a fourth sample data set for the second sample time period, and a fifth sample data set for the third sample time period, the fourth sample data set being a prediction result based on the first sample data set, the fifth sample data set being a prediction result based on the second sample data set, the fourth sample data set comprising: fourth sample data corresponding to the plurality of second sample moments, the fifth sample data set comprising: fifth sample data corresponding to the plurality of third sample moments; training a preset autoregressive conditional heteroscedastic model by using a preset sample residual data cluster, wherein the preset sample residual data cluster comprises: a first set of residual data for a second sample time period, and a second set of residual data for a third sample time period, the first set of residual data comprising: first residual data corresponding to a plurality of second sample moments, wherein the first residual data is a difference value between fourth sample data and second sample data at the same second sample moment, and the second residual data set comprises: and the second residual data corresponds to the plurality of third sample moments, and the second residual data is the difference value between the fifth sample data and the third sample data at the same third sample moment.
In the above embodiment of the present invention, in the process of presetting a prediction model, the preset autoregressive moving average model is used to represent the time-dependent change relationship of data, and the preset autoregressive conditional variance model is used to represent the time-dependent change relationship of random error data of the prediction data, and training data is selected from a preset sample data set, so that the preset autoregressive moving average model can be trained; the prediction data can be obtained based on the trained preset autoregressive moving average model, then real data corresponding to the prediction data can be obtained from the preset sample data cluster, residual data of the prediction data and the real data are based on residual data of the prediction data, the residual data is random error data of the prediction data, the error data is further used as data, a preset autoregressive condition heteroscedastic model can be trained, the error data corresponding to each prediction data can be predicted based on the preset autoregressive condition heteroscedastic model, and accordingly the prediction data of the preset autoregressive moving average model can be corrected based on the predicted error data, and a more accurate prediction result is obtained.
As an alternative embodiment, using the first sample data cluster, training the preset autoregressive moving average model comprises: analyzing the first sample data set using an initial autoregressive moving average model to obtain a sixth sample data set of the second sample time period, wherein the sixth sample data set comprises: sixth sample data corresponding to the second sample moments; comparing the difference between the sixth sample data set and the second sample data set to obtain a first comparison result; and adjusting a first model parameter set in the initial autoregressive moving average model according to the first comparison result to obtain a preset autoregressive moving average model.
According to the embodiment of the invention, the process of training the preset autoregressive moving average model is a process of adjusting a first model parameter set in the model, wherein the first model parameter set can be adjusted according to the prediction data output by the model and the difference between the first model parameter set and the real data, the smaller the difference is, the smaller the model fit is, so that the first model parameter set which enables the model to be better fitted is determined, and then the determined first model parameter set is input into the initial autoregressive moving average model, so that the training of the preset autoregressive moving average model can be completed.
Alternatively, the autoregressive moving average model ARIMA is: ARIMA (p, d, q),wherein p is an autoregressive term, d is a differential number and q is a moving average term, which can be determined empirically; b is denoted as the fallback operator: b (B) k X t ＝X t-k The method comprises the steps of carrying out a first treatment on the surface of the Phi is a parameter in an autoregressive model AR, and represents the influence degree of historical data of an ith time period on current data in the autoregressive model, and is usually estimated by a statistical method such as a least square method; θi is a parameter in the moving average model MA, and represents the degree of influence of the error of the ith time period on the current data in the moving average model, and is usually estimated by a statistical method such as a least square method.
Optionally, the first set of model parameters includes at least: phi, and thetai.
As an alternative embodiment, using the preset sample residual data cluster, training the preset autoregressive conditional covariance model comprises: determining residual errors of the second sample data cluster and the first sample data cluster to obtain a preset sample residual error data cluster; detecting whether preset residual data in a preset sample residual data cluster accords with white noise distribution, wherein the preset residual data at least comprises: first residual data and second residual data; under the condition that the preset residual data accords with white noise distribution, training a preset autoregressive condition heteroscedastic model by using a preset sample residual data set.
According to the embodiment of the invention, the data has the contingency, the complete fitting can not be carried out through the preset autoregressive moving average model, so that the predicted data and the real data can have small errors, the errors can be expressed as white noise, further, after the preset autoregressive moving average model is trained, white noise checking is needed to be carried out based on the predicted data of the preset autoregressive moving average model, if the predicted data accords with the white noise distribution, the predicted data of the preset autoregressive moving average model is accurate enough, and further, the training of the autoregressive conditional heteroscedastic model can be carried out based on the difference value between the predicted data and the real data.
As an alternative embodiment, using the preset sample residual data cluster, training the preset autoregressive conditional covariance model comprises: analyzing the first residual data set by using an initial autoregressive conditional heteroscedastic model to obtain a third residual data set of a third sample time period, wherein the third residual data set comprises: third residual data corresponding to a plurality of third sample moments; comparing the difference between the third residual data set and the second residual data set to obtain a second comparison result; and adjusting a second model parameter set in the initial autoregressive condition heteroscedastic model according to the second comparison result to obtain a preset autoregressive condition heteroscedastic model.
According to the embodiment of the invention, the process of training the preset autoregressive condition heteroscedastic model is a process of adjusting a second model parameter set in the model, wherein the second model parameter set can be adjusted according to the prediction residual data output by the model and the difference between the prediction data and the real residual data of the real data, the smaller the difference is, the smaller the model fit is, the second model parameter set which enables the model to be better fitted is further determined, and then the determined second model parameter set is input into the initial autoregressive condition heteroscedastic model, so that the training of the preset autoregressive condition heteroscedastic model can be completed.
Optionally, the autoregressive conditional heteroscedastic model GARCH is: GARCH (P, Q), σt 2 ＝ω+∑αiεt-i 2 +∑βiσt-i 2 Wherein P is an autocorrelation term for measuring ε t-i 2 The number of epochs (i.e., past volatility information); q is a moving average term used to measure σt-i 2 The number of epochs (i.e., the volatility information of past prediction errors); omega is a constant term; εt is the residual, i.e., the difference between the actual data and the predicted data; sigma t 2 Is the fluctuation rate or variance of time t; αi represents the influence degree of the residual square of the ith time period on the current fluctuation rate in the GARCH model, and is estimated based on a maximum likelihood estimation method; βi represents the degree of influence of the fluctuation rate of the ith time period on the current fluctuation rate in the GARCH model, and is also estimated by the maximum likelihood estimation method.
Optionally, the second set of model parameters includes: αi and βi.
As an alternative embodiment, monitoring the degree of data fluctuation of the predicted data set relative to the historical data set comprises: determining residues of the prediction data set and the historical data set to obtain a prediction residual sequence, wherein the prediction residual sequence comprises: the method comprises the steps of arranging a plurality of prediction residual data according to a time sequence, wherein each prediction residual data is the residual of a predicted first prediction data and a preset historical data at a target moment; determining a statistical index of the predicted residual sequence, wherein the statistical index at least comprises: standard deviation, variance and mean; and evaluating the fluctuation degree of the data according to the statistical index.
According to the embodiment of the invention, the residual errors of the prediction data set and the historical data set are used for carrying out data fluctuation analysis, and the uncertainty of prediction and the fluctuation condition of future data can be known by analyzing the fluctuation of the prediction residual error sequences of the prediction data set and the historical data set.
As an alternative embodiment, after monitoring the degree of data fluctuation in the predicted data set and in the actual data set for the target time period, the method further comprises: displaying the fluctuation degree of the data by using a visual chart, wherein the visual chart at least comprises: graph, report, dashboard, thermodynamic and dynamic charts.
According to the embodiment of the invention, the data fluctuation degree can be displayed and analyzed through the visual chart to evaluate the accuracy of the prediction result and the fluctuation condition of future data.
It should be noted that, a smaller residual fluctuation and a mean value close to zero indicate that the prediction result is more accurate, and a larger residual fluctuation and a mean value far from zero indicate that the prediction result uncertainty is higher.
The invention also provides an alternative embodiment, which provides a fluctuation analysis early warning method based on big data and artificial intelligence technology, and applies the big data and the artificial intelligence technology to the fluctuation analysis early warning method to realize more accurate and more timely monitoring and early warning of fluctuation. The early warning method can be applied to various fields, such as financial markets, environmental monitoring, transportation logistics and the like, and has important application value and market potential.
In recent years, with the rapid development of big data and artificial intelligence technologies, a wave analysis early warning method can be improved and promoted by means of the technologies. The big data technology can collect, process and analyze mass data in real time, and provide more accurate and comprehensive fluctuation information; the artificial intelligence technology can mine potential rules and trends from the data through algorithms such as machine learning, deep learning and the like, and intelligent prediction and early warning of fluctuation conditions are realized.
At present, although big data has been widely used in various fields, there are still some problems in analyzing and utilizing big data. For example, how to analyze massive data rapidly and accurately, and to predict fluctuation and pre-warn through analysis results. The present invention aims to provide an effective method and system that uses big data and artificial intelligence techniques to solve the above problems, help users take measures in time, and reduce possible risks and losses.
Fig. 2 is a schematic diagram of a wave analysis early warning method according to an embodiment of the invention, as shown in fig. 2, including the following steps:
step S202, data collection and preprocessing.
And S204, model construction and training.
Step S206, wave motion recognition and early warning.
Step S208, visualization and analysis.
In the above step S202, the data collection and preprocessing process needs to collect and sort the data to be monitored, and then preprocessing the data to ensure the accuracy and reliability of the analysis result.
Optionally, the data to be monitored includes: sales, access volume, and user behavior, etc.
Optionally, the pretreatment process comprises: data cleaning, outlier removal, data smoothing, and the like.
As an alternative example, the data collection process includes the steps of:
in step S2021, the data type and source to be monitored are determined, for example, sales data may be from an e-commerce platform and access data may be from a website analysis tool.
In step S2022, a data collection system is established, and data is obtained in real time through an API interface, log recording, sensors, etc., so as to ensure that the data can be collected timely and completely.
In step S2023, a storage technique such as a relational database or a distributed file system is selected and used to ensure the security and reliability of the data.
As an alternative example, the data preprocessing procedure includes the steps of:
in step S2026, the data is cleaned, so as to perform cleaning operations such as duplicate value removal and missing value processing on the collected data, so as to ensure the integrity and accuracy of the data.
Step S2026, detecting an abnormal value, which is used to detect the abnormal value of the data, analyze whether the data has an extreme value or an outlier of the abnormality, and process the data according to the requirement. This step may be skipped if the data has already undergone valid outlier processing.
In step S2026, the data is smoothed, and the data with large fluctuation is smoothed, so as to eliminate the influence of random noise on the analysis result. Common smoothing methods include moving average, exponential smoothing, and the like.
The embodiment of the invention aims to ensure the quality and the integrity of data in the data collection and preprocessing stage and provides a reliable data basis for subsequent fluctuation analysis and early warning.
It should be noted that, in some application scenarios, data cleaning and abnormal value removal are not required, including the following application scenarios:
optionally, in the context of data acquisition by a real-time sensor, such as in environmental monitoring, environmental data is acquired in real-time by a sensor device, and because the device itself has high accuracy and reliability, the data is calibrated and screened, generally without obvious outliers or noise, and therefore, no data cleaning and outlier removal process is required.
Optionally, for scenarios where data is exported to a database: the data comes from database systems that have been tightly managed and quality controlled and employ efficient data constraint and validation mechanisms, which have typically eliminated outliers and erroneous data, so that in this case no data cleansing and outlier removal operations are required.
Alternatively, for a scenario where the reliability of the data source is high, if the data source is reliable, the data collection and arrangement process is standardized and is effectively monitored, the probability of outliers and erroneous data in the data is low, and the data quality can be considered to be high enough that no cleaning or outlier removal operations are required.
For example: in the case of dual 11 service high points, data outlier handling may be considered optional, because in this particular scenario, the occurrence of outliers may be normal service behavior, rather than truly outliers.
In the case of the dual 11 service high point, the data outlier processing is not required, and the peak time characteristic and the service requirement are affected.
For example, for peak hour characteristics, double 11 is an important shopping promotion activity of the e-commerce industry during which fluctuations in data and occurrence of outliers are normal, which is caused by an increase in a large amount of user activity and transaction amount. Therefore, some outliers are regarded as normal fluctuations, and no additional processing is required for these outliers.
For another example, during double 11, the e-commerce enterprise may perform some special popularization measures according to the service policy, such as flash purchase, robbery purchase, etc., which may cause abnormal values of data to appear. However, these outliers are set for attracting and exciting the user and need not be considered as data errors or anomalies.
In the above step S204, the model is constructed and trained, and after the data preprocessing is completed, the model is constructed and trained.
Alternatively, the method of combining the GARCH model with the ARIMA model can improve the traditional ARIMA model to have the capability of processing high-frequency data and considering the problem of heteroscedasticity, and the model is called as ARIMA-GARCH model; in time series, characteristics such as fluctuation aggregation and leverage effect can be described using the GARCH model.
FIG. 3 is a schematic diagram of an ARIMA-GARCH model training process according to an embodiment of the present invention, and as shown in FIG. 3, the combination of the ARIMA model and the GARCH model is called ARIMA-GARCH model, which is a common tool for time series analysis in the face of high frequency data and volatility problems, and the specific training process includes the steps of:
step S31, data preprocessing, namely cleaning the data, and processing missing values, abnormal values and the like; and, whether the data is stable or not is determined, if the sequence is not stable, the data can be processed through differential, logarithmic transformation and the like.
Step S32, determining the ARIMA model order, constructing an ARIMA model, and determining optimal parameters, such as an autoregressive term p, a difference frequency d and a moving average term q. The determination can be made by combining the criteria of AIC, BIC and the like through an ACF and a PACF, or an automatic model selection tool can be used.
And step S33, training an ARIMA model, and carrying out parameter training by using a maximum likelihood estimation method or other adaptation methods to obtain parameters of the ARIMA model.
And step S34, residual error checking. After ARIMA model training, residual data are calculated, distribution conditions of the residual data are observed, and model diagnosis is carried out; if the residual data shows white noise distribution, the ARIMA model is well fitted, then the GARCH model is trained, and if the residual data does not meet the white noise distribution, the parameters of the ARIMA model need to be adjusted for retraining.
In step S35, the GARCH model order is determined, and the GARCH model may be constructed based on residual data based on the ARIMA model. The best autocorrelation term P and moving average term Q parameters are determined by combining the autocorrelation function diagram ACF and the partial autocorrelation function diagram PACF with AIC, BIC, etc. criteria.
And step S36, training the GARCH model, and training the GARCH model by using a maximum likelihood estimation method to obtain parameters of the GARCH model.
Step S37, model checking, namely checking the ARIMA-GARCH model overall, and if the ARIMA-GARCH model passes, completing the establishment of the ARIMA-GARCH model; if the test fails, the parameters of the ARIMA model and the parameters of the GARCH model are readjusted for retraining.
Optionally, in the field of big data fluctuation analysis, the ARIMA-GARCH model can be used for early warning risk, such as price fluctuation, flow fluctuation and other risk early warning of stock market, and provides valuable decision support for decision makers. And applying the ARIMA-GARCH model to actual data, performing early warning analysis, and sending out an early warning signal when the predicted fluctuation degree or risk value reaches a fluctuation threshold value.
FIG. 4 is a schematic illustration of training an ARIMA model according to an embodiment of the present invention, as shown in FIG. 4, comprising the steps of:
s41, constructing an ARIMA model.
S42, model training results are 80%.
S43, introducing 20% of expert experience.
S44, obtaining a data predicted value in a future period of time through a prediction function of the model, such as determining a second predicted data set of the target period of time.
S45, fluctuation analysis is performed using the difference (i.e., residual) between the history data and the predicted value.
S46, obtaining a fluctuation early warning analysis result.
Alternatively, in performing model training and wave analysis, a personal expert experience (20% of the ratio) may be used to select an appropriate time series model based on experience and domain knowledge. And carrying out preliminary estimation on the model parameters according to the data characteristics and the historical trend. Reasonable model assumptions and adjustments are made for the particular domain and data characteristics.
Alternatively, in performing the prediction and fluctuation analysis, a predicted value (e.g., a second predicted data set for determining a target time period) in a future period may be calculated based on the historical data and the trained model, and the fluctuation of the future data may be analyzed by calculating a confidence interval for the predicted value (i.e., the second predicted data).
Optionally, during analysis, the ratio of the experience of the personal expert to the training result of the model may be adjusted according to the industry characteristics, and the specific adjustment method may refer to the following steps:
step S2041, industry background analysis. By studying and knowing the characteristics of the target industry, the role of data analysis and expert experience in the industry is known. Knowledge of which questions or tasks are more dependent on expert expertise and which questions or tasks are more dependent on data analysis of the model.
Step S2043, adjusting the weights. And adjusting the expert experience and the weight of the model according to the industry characteristics and task requirements. As much as possible, the sum of the weights of the two is 100%.
Step S2044, implement and feedback. And (3) performing adjusted analysis, and performing further adjustment and optimization by combining actual result feedback.
According to the embodiment of the invention, expert experience and model analysis have advantages, and the advantages of the expert experience and the model analysis are reasonably utilized to effectively analyze and make decisions.
As an alternative embodiment, the prediction data for a future period of time is obtained by a prediction function of the model using the parameters of the trained ARIMA-GARCH model, and the history data as inputs.
Alternatively, there have been trained ARIMA models and GARCH models, with parameters phi i, thetai, alpha i, beta i, respectively. Let P and Q be the autoregressive and moving average terms of the ARIMA model, respectively, and P and Q be the autocorrelation and moving average terms of the GARCH model, respectively. According to the model, prediction data of a future period can be obtained.
Alternatively, the formula of the ARIMA model is written as:
where X is the observed value (e.g., historical data) and ε is the random error data.
Alternatively, the formula of the GARCH model is written as: epsilon t ＝σ t *Z t ，
Wherein Z is t -iid N (0, 1) is a standard normal distribution.
Optionally, aBy combining the ARIMA model and the GARCH model, the observation value (such as prediction data) X of the next period can be predicted {t+1 }。
Wherein ε {t+1} ＝σ {t+1} *Z {t+1} ，
Alternatively, the parameters obtained by the ARIMA-GARCH module and the known historical data are substituted into the formula to obtain the predicted value (i.e., determine the predicted data) of the future period.
Because of the characteristics of the GARCH model, uncertainty in the estimation error needs to be considered at the same time when making predictions.
Alternatively, future prediction data may be obtained and further analyzed based on the prediction of the ARIMA-GARCH module.
In step S206 described above, once the model has been trained, it may be applied to a real-time data monitoring process. In the monitoring process, the model can carry out fluctuation recognition and early warning judgment on the current data. If the fluctuation degree of the data exceeds a preset fluctuation threshold, the system automatically triggers an early warning mechanism and sends an alarm or notification to related personnel so as to take measures in time.
As an alternative example, the wave analysis includes the steps of:
step S2061, performing fluctuation analysis using the difference (i.e., residual) between the historical data and the predicted data, and calculating a residual sequence res=y-y of the historical data and model fitting pred Where y is historical data, y pred Is the corresponding prediction data.
In step S2062, the fluctuation of the residual sequence is analyzed to obtain the uncertainty of the prediction and the fluctuation of the future data. The analysis method adopts statistical indexes such as standard deviation, variance, mean value and the like of a calculated residual sequence, and draws a residual image and an autocorrelation image.
In step S2063, the accuracy of prediction and the fluctuation of future data are evaluated according to the statistical index of the residual error and the graph analysis result, wherein smaller residual error fluctuation and the mean value close to zero represent that the prediction is more accurate, and larger residual error fluctuation and the mean value far from zero represent that the prediction uncertainty is higher.
It should be noted that, the residual sequence of the data may be analyzed, and the fluctuation degree of the current data may be determined by using statistical methods such as standard deviation, variance, etc., or according to the distribution condition of the historical data. And comparing the fluctuation degree of the real-time data with the fluctuation threshold according to the preset fluctuation threshold, and judging whether to trigger fluctuation early warning.
In the process of carrying out fluctuation early warning based on big data, the residual error is analyzed on the historical data, the model parameters or expert experience indexes are continuously optimized, the noise reduction effect is achieved, and the fluctuation early warning result can be more accurate.
As an alternative example, the noise reduction process includes the steps of:
in step S2064, data is collected and consolidated. By collecting the predicted and actual data of the model, the same timeline and other corresponding labels are used for sorting in order to accurately calculate the residual.
In step S2065, a residual error is calculated.
The residual error refers to a difference between the predicted data and the actual data, and the specific calculation method is to use an actual value to subtract the predicted value from each item of data.
In step S2066, the residual distribution is observed. The distribution form of the residual is observed by plotting a distribution map (such as a histogram, a box chart and the like) of the residual.
Alternatively, if the residual is concentrated near zero and the morphology is approximately normal, indicating that the predictive performance of the model is relatively good, the noise is mainly from random factors; if the residual exhibits some systematic pattern, there may be bias to the model.
Step S2067, residual sequence detection. By carrying out time sequence diagram analysis on the residual sequence, whether seasonality, trend and the like exist can be judged; an autocorrelation check may also be performed to determine if there is a correlation between the previous and subsequent values of the residual.
In step S2068, the noise source is analyzed. By identifying possible sources of noise, it is determined whether the noise is from a data quality problem (e.g., erroneous data, outliers, etc.), a model problem (e.g., model bias, over-fit, or under-fit), or other cause (e.g., environmental changes, incidents occurring during prediction, etc.).
Step S2069, processing noise. And according to the analysis result of the noise source, adopting a corresponding method to treat the corresponding problem. For example, data errors are repaired, models are optimized, noise reduction methods such as filters are adopted, and the like.
It should be noted that, steps S2064 to 2069 are iterative processes, and each time the noise is processed, the residual error needs to be recalculated and analyzed until a satisfactory effect is achieved.
As an alternative example, if the degree of fluctuation of the data exceeds a pre-set pre-alarm threshold, the system will automatically trigger the fluctuation pre-alarm mechanism.
Optionally, the early warning mechanism may transmit early warning information to related personnel by sending an alarm, a short message, an email or other notification means; and then related personnel can immediately take corresponding measures or decisions according to the received early warning information so as to cope with the fluctuation condition.
Optionally, in the wave identification and early warning process, the wave threshold value can be dynamically adjusted according to specific conditions, so that the early warning mechanism is more accurate and reliable.
Optionally, in the wave analysis and early warning process, other factors such as external environment factors, market changes and the like can be combined to perform comprehensive wave analysis and early warning judgment. In this way, abnormal fluctuations can be found in time and appropriate measures can be taken to reduce potential risks and losses.
In the step S208, in order to facilitate the user to understand and analyze the fluctuation of the data, the monitoring result is displayed to the user through the visual means, such as graph, report, dashboard, thermodynamic diagram, dynamic chart, etc. The visualization can intuitively display the fluctuation condition of the data, help a user to quickly identify and analyze the problem, and make corresponding decisions and adjustments.
Alternatively, the historical data and the predictive data may be presented in continuous lines using a graph. Different colors or line types may be used to distinguish between observed and predicted data in order to intuitively identify fluctuations and trend changes in the data.
Alternatively, the report may contain various statistical indicators and trends in the change in the indicators. The method is displayed in forms of a table, a chart, a pie chart and the like, so that a user can quickly know the fluctuation condition of data and the change of key indexes.
Optionally, the instrument panel is a visualization tool integrated with multiple indexes, and can display fluctuation conditions of multiple key indexes at one time. The user can easily monitor a plurality of indexes through the instrument panel and make decisions according to the fluctuation condition.
Alternatively, the thermodynamic diagram shows the data fluctuations for different time periods or different regions in a color coded manner. The intensity and fluctuation degree of the data can be indicated according to the color shade, so that a user can be helped to intuitively find abnormal fluctuation in a large amount of data.
Alternatively, the fluctuation of the data can be displayed in a time series manner using a dynamic graph, such as a scroll graph or an animation graph. This form of visualization enables the user to better observe the fluctuating trends and changes in the data.
According to the embodiment of the invention, through a visual means, a user can more conveniently understand and analyze the fluctuation condition of data, quickly identify abnormal fluctuation and make corresponding decisions and adjustments; at the same time, the visualization is also helpful for finding hidden patterns and associations in the data, and provides guidance for further analysis and prediction.
The technical scheme provided by the invention has high accuracy, and through the application of big data analysis and artificial intelligence technology, the fluctuation conditions in different fields can be accurately mastered, the accuracy of prediction is improved, and a user is helped to make a more intelligent decision.
The technical scheme provided by the invention has high efficiency, adopts a distributed computing and storing technology to process big data, improves the processing speed and efficiency, ensures that the result can be obtained in the shortest time by prediction and early warning, and provides timely information support for users; the method has strong adaptability, can be suitable for fluctuation analysis and early warning in various fields, has strong adaptability and flexibility, and can meet the requirements of different industries and fields; the intelligent support is provided, besides providing accurate prediction and early warning information, intelligent decision support can be provided for the user according to the prediction result, the user can be helped to make a correct decision, measures can be taken in time, risks are prevented, and loss is reduced; the labor cost can be reduced, the fluctuation analysis and the early warning task can be completed under the condition of less human participation by utilizing automated big data and artificial intelligence technology, and the labor cost and the working strength are reduced.
In conclusion, the technical scheme provided by the application can provide accurate, efficient and intelligent fluctuation analysis early warning service, helps users reduce risks and losses, improves working efficiency and has wide application prospects.
According to the embodiment of the invention, a data fluctuation early-warning device embodiment is provided, and it is to be noted that the data fluctuation early-warning device can be used for executing the data fluctuation early-warning method in the embodiment of the invention, and the data fluctuation early-warning method in the embodiment of the invention can be executed in the data fluctuation early-warning device.
Fig. 5 is a schematic diagram of a data fluctuation pre-warning device according to an embodiment of the present invention, as shown in fig. 5, the device may include: an obtaining module 52, configured to obtain a historical dataset of historical time periods, where the historical time periods include: the plurality of historical moments, the historical dataset comprising: historical data corresponding to a plurality of historical moments; the analysis module 54 is configured to analyze the historical data set using a preset prediction model to obtain a first predicted data set of the target time period, where the preset prediction model at least includes: the method comprises the steps of presetting an autoregressive moving average model and a preset autoregressive condition heteroscedastic model, wherein the preset autoregressive moving average model is used for predicting a second prediction data set of a target time period according to a historical data set, the preset autoregressive condition heteroscedastic model is used for predicting an error data set of the target time period, and the target time period comprises: the plurality of target moments, the first prediction dataset comprising: the first predicted data corresponding to the at least one target time, the second predicted data set comprising: the second prediction data corresponding to the at least one target moment, and the error data set comprises: at least one error data corresponding to the target moment, wherein the first predicted data is the correction result of the second predicted data and the error data of the same target moment; a monitoring module 56 for monitoring the degree of data fluctuation of the predicted data set relative to the historical data set; and the early warning module 58 is used for carrying out fluctuation early warning under the condition that the fluctuation degree of the data exceeds a fluctuation threshold value.
It should be noted that, the acquiring module 52 in this embodiment may be used to perform step S102 in the embodiment of the present application, the analyzing module 54 in this embodiment may be used to perform step S104 in the embodiment of the present application, the monitoring module 56 in this embodiment may be used to perform step S106 in the embodiment of the present application, and the early warning module 58 in this embodiment may be used to perform step S108 in the embodiment of the present application. The above modules are the same as examples and application scenarios implemented by the corresponding steps, but are not limited to what is disclosed in the above embodiments.
In the embodiment of the invention, the preset prediction model can be determined by utilizing an artificial intelligence technology through algorithms such as machine learning, deep learning and the like and is used for mining potential rules and trends from data, and the preset prediction model combining the preset autoregressive moving average model and the preset autoregressive conditional heteroscedure model has the capability of processing high-frequency data, considering the problem of heteroscedasticity and the like, the second preset data set of a target time period can be predicted according to a historical data set through the preset autoregressive moving average model, the error uncertainty of a prediction result can be considered through the preset autoregressive conditional heteroscedure model, the second preset data set can be corrected, a more accurate second prediction data set can be obtained, further, the data fluctuation degree of the target time period can be predicted based on the second prediction data set, and early warning can be carried out according to the predicted data fluctuation degree, so that the technical effects of intelligent prediction and early warning of fluctuation conditions are realized, and the technical problem that the prior art cannot carry out data fluctuation analysis on a large-scale data environment is solved.
As an alternative embodiment, the apparatus further comprises: the acquisition sub-module is used for acquiring a preset sample data set group of a preset sample time period before analyzing the historical data set by using a preset prediction model to obtain a first prediction data set of a target time period, wherein the preset sample time period at least comprises a first sample time period, a second sample time period and a third sample time period which are continuous, and the first sample time period comprises: the plurality of first sample moments, the second sample time period comprising: the plurality of second sample times, the third sample time period comprising: the preset sample data clusters at least comprise a plurality of third sample moments: a first sample data set for a first sample time period, a second sample data set for a second sample time period, and a third sample data set for a third sample time period, the first sample data set comprising: the first sample data corresponding to the plurality of first sample moments, the second sample data set including: the second sample data corresponding to the plurality of second sample moments, and the third sample data set includes: third sample data corresponding to a plurality of third sample moments; a first training sub-module for training a preset autoregressive moving average model using a first sample data cluster, wherein the first sample data cluster comprises: a first sample data set and a second sample data set; the analysis submodule is used for analyzing the first sample data cluster by using a preset autoregressive moving average model to obtain a second sample data cluster, wherein the second sample data cluster comprises: a fourth sample data set for the second sample time period, and a fifth sample data set for the third sample time period, the fourth sample data set being a prediction result based on the first sample data set, the fifth sample data set being a prediction result based on the second sample data set, the fourth sample data set comprising: fourth sample data corresponding to the plurality of second sample moments, the fifth sample data set comprising: fifth sample data corresponding to the plurality of third sample moments; the second training submodule is used for training a preset autoregressive conditional heteroscedastic model by using a preset sample residual data cluster, wherein the preset sample residual data cluster comprises: a first set of residual data for a second sample time period, and a second set of residual data for a third sample time period, the first set of residual data comprising: first residual data corresponding to a plurality of second sample moments, wherein the first residual data is a difference value between fourth sample data and second sample data at the same second sample moment, and the second residual data set comprises: and the second residual data corresponds to the plurality of third sample moments, and the second residual data is the difference value between the fifth sample data and the third sample data at the same third sample moment.
As an alternative embodiment, the first training submodule comprises: a first analysis unit, configured to analyze the first sample data set using an initial autoregressive moving average model, to obtain a sixth sample data set of the second sample time period, where the sixth sample data set includes: sixth sample data corresponding to the second sample moments; the first comparison unit is used for comparing the difference between the sixth sample data set and the second sample data set to obtain a first comparison result; and the first adjusting unit is used for adjusting a first model parameter set in the initial autoregressive moving average model according to the first comparison result to obtain a preset autoregressive moving average model.
As an alternative embodiment, the second training submodule comprises: the determining unit is used for determining residual errors of the second sample data cluster and the first sample data cluster to obtain a preset sample residual error data cluster; the detection unit is used for detecting whether preset residual data in the preset sample residual data cluster accords with white noise distribution, wherein the preset residual data at least comprises: first residual data and second residual data; and the training unit is used for training a preset autoregressive condition heteroscedastic model by using the preset sample residual data set under the condition that the preset residual data accords with the white noise distribution.
As an alternative embodiment, the second training submodule comprises: the second analysis unit is configured to analyze the first residual data set by using an initial autoregressive conditional heteroscedastic model to obtain a third residual data set in a third sample time period, where the third residual data set includes: third residual data corresponding to a plurality of third sample moments; the second comparison unit is used for comparing the difference between the third residual data set and the second residual data set to obtain a second comparison result; and the second adjusting unit is used for adjusting a second model parameter set in the initial autoregressive condition heteroscedastic model according to a second comparison result to obtain a preset autoregressive condition heteroscedastic model.
As an alternative embodiment, the monitoring module comprises: a first determining subunit, configured to determine a residual error between the predicted data set and the historical data set, to obtain a predicted residual error sequence, where the predicted residual error sequence includes: the method comprises the steps of arranging a plurality of prediction residual data according to a time sequence, wherein each prediction residual data is the residual of a predicted first prediction data and a preset historical data at a target moment; a second determining subunit, configured to determine a statistical indicator of the prediction residual sequence, where the statistical indicator at least includes: standard deviation, variance and mean; and the evaluation subunit is used for evaluating the fluctuation degree of the data according to the statistical indexes.
As an alternative embodiment, the apparatus further comprises: and the display submodule is used for displaying the data fluctuation degree by using a visual chart after monitoring the data fluctuation degree in the actual data set of the prediction data set and the target time period, wherein the visual chart at least comprises: graph, report, dashboard, thermodynamic and dynamic charts.
Embodiments of the present invention may provide a computer terminal, which may be any one of a group of computer terminals. Alternatively, in the present embodiment, the above-described computer terminal may be replaced with a terminal device such as a mobile terminal.
Alternatively, in this embodiment, the above-mentioned computer terminal may be located in at least one network device among a plurality of network devices of the computer network.
In this embodiment, the computer terminal may execute the program code of the following steps in the method for early warning fluctuation of data: a historical dataset of historical time periods is obtained, wherein the historical time periods include: the plurality of historical moments, the historical dataset comprising: historical data corresponding to a plurality of historical moments; analyzing the historical data set by using a preset prediction model to obtain a first prediction data set of a target time period, wherein the preset prediction model at least comprises: the method comprises the steps of presetting an autoregressive moving average model and a preset autoregressive condition heteroscedastic model, wherein the preset autoregressive moving average model is used for predicting a second prediction data set of a target time period according to a historical data set, the preset autoregressive condition heteroscedastic model is used for predicting an error data set of the target time period, and the target time period comprises: the plurality of target moments, the first prediction dataset comprising: the first predicted data corresponding to the at least one target time, the second predicted data set comprising: the second prediction data corresponding to the at least one target moment, and the error data set comprises: at least one error data corresponding to the target moment, wherein the first predicted data is the correction result of the second predicted data and the error data of the same target moment; monitoring the degree of data fluctuation of the predicted dataset relative to the historical dataset; and carrying out fluctuation early warning under the condition that the fluctuation degree of the data exceeds a fluctuation threshold value.
Alternatively, fig. 6 is a block diagram of a computer terminal according to an embodiment of the present invention. As shown in fig. 6, the computer terminal 60 may include: one or more (only one is shown) processors 62, and memory 64.
The memory may be used to store software programs and modules, such as program instructions/modules corresponding to the method and apparatus for early warning fluctuation of data in the embodiments of the present invention, and the processor executes the software programs and modules stored in the memory, thereby executing various functional applications and data processing, that is, implementing the method for early warning fluctuation of data. The memory may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory may further include memory remotely located with respect to the processor, which may be connected to the terminal 60 via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
The processor may call the information and the application program stored in the memory through the transmission device to perform the following steps: a historical dataset of historical time periods is obtained, wherein the historical time periods include: the plurality of historical moments, the historical dataset comprising: historical data corresponding to a plurality of historical moments; analyzing the historical data set by using a preset prediction model to obtain a first prediction data set of a target time period, wherein the preset prediction model at least comprises: the method comprises the steps of presetting an autoregressive moving average model and a preset autoregressive condition heteroscedastic model, wherein the preset autoregressive moving average model is used for predicting a second prediction data set of a target time period according to a historical data set, the preset autoregressive condition heteroscedastic model is used for predicting an error data set of the target time period, and the target time period comprises: the plurality of target moments, the first prediction dataset comprising: the first predicted data corresponding to the at least one target time, the second predicted data set comprising: the second prediction data corresponding to the at least one target moment, and the error data set comprises: at least one error data corresponding to the target moment, wherein the first predicted data is the correction result of the second predicted data and the error data of the same target moment; monitoring the degree of data fluctuation of the predicted dataset relative to the historical dataset; and carrying out fluctuation early warning under the condition that the fluctuation degree of the data exceeds a fluctuation threshold value.
Optionally, the above processor may further execute program code for: before analyzing the historical data set by using a preset prediction model to obtain a first prediction data set of a target time period, obtaining a preset sample data set of a preset sample time period, wherein the preset sample time period at least comprises a first sample time period, a second sample time period and a third sample time period which are continuous, and the first sample time period comprises: the plurality of first sample moments, the second sample time period comprising: the plurality of second sample times, the third sample time period comprising: the preset sample data clusters at least comprise a plurality of third sample moments: a first sample data set for a first sample time period, a second sample data set for a second sample time period, and a third sample data set for a third sample time period, the first sample data set comprising: the first sample data corresponding to the plurality of first sample moments, the second sample data set including: the second sample data corresponding to the plurality of second sample moments, and the third sample data set includes: third sample data corresponding to a plurality of third sample moments; training a preset autoregressive moving average model using a first sample data cluster, wherein the first sample data cluster comprises: a first sample data set and a second sample data set; analyzing the first sample data cluster by using a preset autoregressive moving average model to obtain a second sample data cluster, wherein the second sample data cluster comprises: a fourth sample data set for the second sample time period, and a fifth sample data set for the third sample time period, the fourth sample data set being a prediction result based on the first sample data set, the fifth sample data set being a prediction result based on the second sample data set, the fourth sample data set comprising: fourth sample data corresponding to the plurality of second sample moments, the fifth sample data set comprising: fifth sample data corresponding to the plurality of third sample moments; training a preset autoregressive conditional heteroscedastic model by using a preset sample residual data cluster, wherein the preset sample residual data cluster comprises: a first set of residual data for a second sample time period, and a second set of residual data for a third sample time period, the first set of residual data comprising: first residual data corresponding to a plurality of second sample moments, wherein the first residual data is a difference value between fourth sample data and second sample data at the same second sample moment, and the second residual data set comprises: and the second residual data corresponds to the plurality of third sample moments, and the second residual data is the difference value between the fifth sample data and the third sample data at the same third sample moment.
Optionally, the above processor may further execute program code for: analyzing the first sample data set using an initial autoregressive moving average model to obtain a sixth sample data set of the second sample time period, wherein the sixth sample data set comprises: sixth sample data corresponding to the second sample moments; comparing the difference between the sixth sample data set and the second sample data set to obtain a first comparison result; and adjusting a first model parameter set in the initial autoregressive moving average model according to the first comparison result to obtain a preset autoregressive moving average model.
Optionally, the above processor may further execute program code for: determining residual errors of the second sample data cluster and the first sample data cluster to obtain a preset sample residual error data cluster; detecting whether preset residual data in a preset sample residual data cluster accords with white noise distribution, wherein the preset residual data at least comprises: first residual data and second residual data; under the condition that the preset residual data accords with white noise distribution, training a preset autoregressive condition heteroscedastic model by using a preset sample residual data set.
Optionally, the above processor may further execute program code for: analyzing the first residual data set by using an initial autoregressive conditional heteroscedastic model to obtain a third residual data set of a third sample time period, wherein the third residual data set comprises: third residual data corresponding to a plurality of third sample moments; comparing the difference between the third residual data set and the second residual data set to obtain a second comparison result; and adjusting a second model parameter set in the initial autoregressive condition heteroscedastic model according to the second comparison result to obtain a preset autoregressive condition heteroscedastic model.
Optionally, the above processor may further execute program code for: determining residues of the prediction data set and the historical data set to obtain a prediction residual sequence, wherein the prediction residual sequence comprises: the method comprises the steps of arranging a plurality of prediction residual data according to a time sequence, wherein each prediction residual data is the residual of a predicted first prediction data and a preset historical data at a target moment; determining a statistical index of the predicted residual sequence, wherein the statistical index at least comprises: standard deviation, variance and mean; and evaluating the fluctuation degree of the data according to the statistical index.
Optionally, the above processor may further execute program code for: after monitoring the degree of data fluctuation in the predicted dataset and the actual dataset for the target time period, displaying the degree of data fluctuation using a visual chart, wherein the visual chart at least comprises: graph, report, dashboard, thermodynamic and dynamic charts.
By adopting the embodiment of the invention, a scheme for early warning the fluctuation of the data is provided. In the embodiment of the invention, the preset prediction model can be determined by utilizing an artificial intelligence technology through algorithms such as machine learning, deep learning and the like and is used for mining potential rules and trends from data, and the preset prediction model combining the preset autoregressive moving average model and the preset autoregressive conditional heteroscedure model has the capability of processing high-frequency data, considering the problem of heteroscedasticity and the like, the second preset data set of a target time period can be predicted according to a historical data set through the preset autoregressive moving average model, the error uncertainty of a prediction result can be considered through the preset autoregressive conditional heteroscedure model, the second preset data set can be corrected, a more accurate second prediction data set can be obtained, further, the data fluctuation degree of the target time period can be predicted based on the second prediction data set, and early warning can be carried out according to the predicted data fluctuation degree, so that the technical effects of intelligent prediction and early warning of fluctuation conditions are realized, and the technical problem that the prior art cannot carry out data fluctuation analysis on a large-scale data environment is solved.
It will be appreciated by those skilled in the art that the configuration shown in fig. 6 is only illustrative, and the computer terminal may be a smart phone (such as an Android phone, an iOS phone, etc.), a tablet computer, a palm-phone computer, a mobile internet device (Mobile Internet Devices, MID), a PAD, etc. Fig. 6 is not limited to the structure of the electronic device. For example, the computer terminal 60 may also include more or fewer components (e.g., network interfaces, display devices, etc.) than shown in FIG. 6, or have a different configuration than shown in FIG. 6.
Those skilled in the art will appreciate that all or part of the steps in the various methods of the above embodiments may be implemented by a program for instructing a terminal device to execute on hardware associated with the terminal device, the program may be stored in a nonvolatile storage medium, and the nonvolatile storage medium may include: flash disk, read-Only Memory (ROM), random-access Memory (Random Access Memory, RAM), magnetic or optical disk, and the like.
Embodiments of the present invention also provide a nonvolatile storage medium. Alternatively, in the present embodiment, the above-described nonvolatile storage medium may be used to store the program code executed by the fluctuation warning method of data provided in the above-described embodiment.
Alternatively, in this embodiment, the above-mentioned nonvolatile storage medium may be located in any one of the computer terminals in the computer terminal group in the computer network, or in any one of the mobile terminals in the mobile terminal group.
Optionally, in the present embodiment, the non-volatile storage medium is arranged to store program code for performing the steps of: a historical dataset of historical time periods is obtained, wherein the historical time periods include: the plurality of historical moments, the historical dataset comprising: historical data corresponding to a plurality of historical moments; analyzing the historical data set by using a preset prediction model to obtain a first prediction data set of a target time period, wherein the preset prediction model at least comprises: the method comprises the steps of presetting an autoregressive moving average model and a preset autoregressive condition heteroscedastic model, wherein the preset autoregressive moving average model is used for predicting a second prediction data set of a target time period according to a historical data set, the preset autoregressive condition heteroscedastic model is used for predicting an error data set of the target time period, and the target time period comprises: the plurality of target moments, the first prediction dataset comprising: the first predicted data corresponding to the at least one target time, the second predicted data set comprising: the second prediction data corresponding to the at least one target moment, and the error data set comprises: at least one error data corresponding to the target moment, wherein the first predicted data is the correction result of the second predicted data and the error data of the same target moment; monitoring the degree of data fluctuation of the predicted dataset relative to the historical dataset; and carrying out fluctuation early warning under the condition that the fluctuation degree of the data exceeds a fluctuation threshold value.
Optionally, in the present embodiment, the non-volatile storage medium is arranged to store program code for performing the steps of: before analyzing the historical data set by using a preset prediction model to obtain a first prediction data set of a target time period, obtaining a preset sample data set of a preset sample time period, wherein the preset sample time period at least comprises a first sample time period, a second sample time period and a third sample time period which are continuous, and the first sample time period comprises: the plurality of first sample moments, the second sample time period comprising: the plurality of second sample times, the third sample time period comprising: the preset sample data clusters at least comprise a plurality of third sample moments: a first sample data set for a first sample time period, a second sample data set for a second sample time period, and a third sample data set for a third sample time period, the first sample data set comprising: the first sample data corresponding to the plurality of first sample moments, the second sample data set including: the second sample data corresponding to the plurality of second sample moments, and the third sample data set includes: third sample data corresponding to a plurality of third sample moments; training a preset autoregressive moving average model using a first sample data cluster, wherein the first sample data cluster comprises: a first sample data set and a second sample data set; analyzing the first sample data cluster by using a preset autoregressive moving average model to obtain a second sample data cluster, wherein the second sample data cluster comprises: a fourth sample data set for the second sample time period, and a fifth sample data set for the third sample time period, the fourth sample data set being a prediction result based on the first sample data set, the fifth sample data set being a prediction result based on the second sample data set, the fourth sample data set comprising: fourth sample data corresponding to the plurality of second sample moments, the fifth sample data set comprising: fifth sample data corresponding to the plurality of third sample moments; training a preset autoregressive conditional heteroscedastic model by using a preset sample residual data cluster, wherein the preset sample residual data cluster comprises: a first set of residual data for a second sample time period, and a second set of residual data for a third sample time period, the first set of residual data comprising: first residual data corresponding to a plurality of second sample moments, wherein the first residual data is a difference value between fourth sample data and second sample data at the same second sample moment, and the second residual data set comprises: and the second residual data corresponds to the plurality of third sample moments, and the second residual data is the difference value between the fifth sample data and the third sample data at the same third sample moment.
Optionally, in the present embodiment, the non-volatile storage medium is arranged to store program code for performing the steps of: analyzing the first sample data set using an initial autoregressive moving average model to obtain a sixth sample data set of the second sample time period, wherein the sixth sample data set comprises: sixth sample data corresponding to the second sample moments; comparing the difference between the sixth sample data set and the second sample data set to obtain a first comparison result; and adjusting a first model parameter set in the initial autoregressive moving average model according to the first comparison result to obtain a preset autoregressive moving average model.
Optionally, in the present embodiment, the non-volatile storage medium is arranged to store program code for performing the steps of: determining residual errors of the second sample data cluster and the first sample data cluster to obtain a preset sample residual error data cluster; detecting whether preset residual data in a preset sample residual data cluster accords with white noise distribution, wherein the preset residual data at least comprises: first residual data and second residual data; under the condition that the preset residual data accords with white noise distribution, training a preset autoregressive condition heteroscedastic model by using a preset sample residual data set.
Optionally, in the present embodiment, the non-volatile storage medium is arranged to store program code for performing the steps of: analyzing the first residual data set by using an initial autoregressive conditional heteroscedastic model to obtain a third residual data set of a third sample time period, wherein the third residual data set comprises: third residual data corresponding to a plurality of third sample moments; comparing the difference between the third residual data set and the second residual data set to obtain a second comparison result; and adjusting a second model parameter set in the initial autoregressive condition heteroscedastic model according to the second comparison result to obtain a preset autoregressive condition heteroscedastic model.
Optionally, in the present embodiment, the non-volatile storage medium is arranged to store program code for performing the steps of: determining residues of the prediction data set and the historical data set to obtain a prediction residual sequence, wherein the prediction residual sequence comprises: the method comprises the steps of arranging a plurality of prediction residual data according to a time sequence, wherein each prediction residual data is the residual of a predicted first prediction data and a preset historical data at a target moment; determining a statistical index of the predicted residual sequence, wherein the statistical index at least comprises: standard deviation, variance and mean; and evaluating the fluctuation degree of the data according to the statistical index.
Optionally, in the present embodiment, the non-volatile storage medium is arranged to store program code for performing the steps of: after monitoring the degree of data fluctuation in the predicted dataset and the actual dataset for the target time period, displaying the degree of data fluctuation using a visual chart, wherein the visual chart at least comprises: graph, report, dashboard, thermodynamic and dynamic charts.
The foregoing embodiment numbers of the present invention are merely for the purpose of description, and do not represent the advantages or disadvantages of the embodiments.
In the foregoing embodiments of the present invention, the descriptions of the embodiments are emphasized, and for a portion of this disclosure that is not described in detail in this embodiment, reference is made to the related descriptions of other embodiments.
In the several embodiments provided in the present application, it should be understood that the disclosed technology content may be implemented in other manners. The above-described embodiments of the apparatus are merely exemplary, and the division of the units, for example, may be a logic function division, and may be implemented in another manner, for example, a plurality of units or components may be combined or may be integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be through some interfaces, units or modules, or may be in electrical or other forms.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional unit in the embodiments of the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units.
The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a non-volatile storage medium. Based on such understanding, the technical solution of the present invention may be embodied essentially or in part or all of the technical solution or in part in the form of a software product stored in a non-volatile storage medium, including instructions for causing a computer device (which may be a personal computer, a server, a network device, or the like) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned nonvolatile storage medium includes: a U-disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a removable hard disk, a magnetic disk, or an optical disk, or other various media capable of storing program codes.
The foregoing is merely a preferred embodiment of the present invention and it should be noted that modifications and adaptations to those skilled in the art may be made without departing from the principles of the present invention, which are intended to be comprehended within the scope of the present invention.
Claims (10)
1. The method for early warning the fluctuation of the data is characterized by comprising the following steps of:
a historical dataset of a historical time period is obtained, wherein the historical time period comprises: a plurality of historical moments, the historical dataset comprising: historical data corresponding to a plurality of historical moments;
analyzing the historical data set by using a preset prediction model to obtain a first prediction data set of a target time period, wherein the preset prediction model at least comprises: the method comprises the steps of presetting an autoregressive moving average model and a preset autoregressive conditional covariance model, wherein the preset autoregressive moving average model is used for predicting a second prediction data set of the target time period according to the historical data set, the preset autoregressive conditional covariance model is used for predicting an error data set of the target time period, and the target time period comprises: a plurality of target moments, the first prediction dataset comprising: at least one first prediction data corresponding to the target moment, wherein the second prediction data set comprises: at least one second prediction data corresponding to the target moment, wherein the error data set comprises: at least one error data corresponding to the target time, wherein the first prediction data is a correction result of the second prediction data and the error data of the same target time;
Monitoring a degree of data fluctuation of the predictive dataset relative to the historical dataset;
and carrying out fluctuation early warning under the condition that the fluctuation degree of the data exceeds a fluctuation threshold value.
2. The method of claim 1, wherein prior to analyzing the historical dataset using a pre-set predictive model to obtain a first predictive dataset for a target time period, the method further comprises:
obtaining a preset sample data cluster of a preset sample time period, wherein the preset sample time period at least comprises a first sample time period, a second sample time period and a third sample time period which are continuous, and the first sample time period comprises: a plurality of first sample times, the second sample time period comprising: a plurality of second sample times, the third sample time period comprising: a plurality of third sample moments, the preset sample data clusters at least comprise: a first sample data set for the first sample time period, a second sample data set for the second sample time period, and a third sample data set for the third sample time period, the first sample data set comprising: a plurality of first sample data corresponding to the first sample time, and the second sample data set includes: and a plurality of second sample data corresponding to the second sample moments, wherein the third sample data set comprises: third sample data corresponding to a plurality of third sample moments;
Training the preset autoregressive moving average model using a first sample data cluster, wherein the first sample data cluster comprises: the first sample data set and the second sample data set;
analyzing the first sample data cluster by using the preset autoregressive moving average model to obtain a second sample data cluster, wherein the second sample data cluster comprises: a fourth sample data set for the second sample time period, the fourth sample data set being a prediction result based on the first sample data set, and a fifth sample data set for the third sample time period, the fifth sample data set being a prediction result based on the second sample data set, the fourth sample data set comprising: fourth sample data corresponding to the second sample moments, wherein the fifth sample data set comprises: fifth sample data corresponding to the third sample moments;
training the preset autoregressive conditional heteroscedastic model using a preset sample residual data cluster, wherein the preset sample residual data cluster comprises: a first set of residual data for the second sample time period and a second set of residual data for the third sample time period, the first set of residual data comprising: first residual data corresponding to a plurality of second sample moments, wherein the first residual data is a difference value between the fourth sample data and the second sample data at the same second sample moment, and the second residual data set comprises: and second residual data corresponding to the third sample time, wherein the second residual data is a difference value between the fifth sample data and the third sample data at the same third sample time.
3. The method of claim 2, wherein training the preset autoregressive moving average model using the first sample data cluster comprises:
analyzing the first sample data set using an initial autoregressive moving average model to obtain a sixth sample data set of the second sample time period, wherein the sixth sample data set comprises: a plurality of sixth sample data corresponding to the second sample time;
comparing the difference between the sixth sample data set and the second sample data set to obtain a first comparison result;
and adjusting a first model parameter set in the initial autoregressive moving average model according to the first comparison result to obtain the preset autoregressive moving average model.
4. The method of claim 2, wherein training the preset autoregressive conditional heteroscedastic model using a preset sample residual data cluster comprises:
determining residual errors of the second sample data cluster and the first sample data cluster to obtain a preset sample residual error data cluster;
detecting whether preset residual data in the preset sample residual data cluster accords with white noise distribution, wherein the preset residual data at least comprises: the first residual data and the second residual data;
And under the condition that the preset residual data accords with the white noise distribution, training the preset autoregressive condition heteroscedastic model by using the preset sample residual data set.
5. The method of claim 2, wherein training the preset autoregressive conditional heteroscedastic model using a preset sample residual data cluster comprises:
analyzing the first residual data set by using an initial autoregressive conditional heteroscedastic model to obtain a third residual data set of the third sample time period, wherein the third residual data set comprises: third residual data corresponding to a plurality of third sample moments;
comparing the difference between the third residual data set and the second residual data set to obtain a second comparison result;
and adjusting a second model parameter set in the initial autoregressive condition heteroscedastic model according to the second comparison result to obtain the preset autoregressive condition heteroscedastic model.
6. The method of claim 1, wherein monitoring the extent of data fluctuation of the predictive dataset relative to the historical dataset comprises:
determining residuals of the prediction data set and the historical data set to obtain a prediction residual sequence, wherein the prediction residual sequence comprises: the prediction residual data are arranged according to a time sequence, and each prediction residual data is the residual of the predicted first prediction data and the preset historical data at the target moment;
Determining a statistical index of the prediction residual sequence, wherein the statistical index at least comprises: standard deviation, variance and mean;
and evaluating the fluctuation degree of the data according to the statistical index.
7. The method of claim 1, wherein after monitoring the extent of data fluctuation in the predicted data set and the actual data set for the target time period, the method further comprises:
displaying the degree of fluctuation of the data by using a visual chart, wherein the visual chart at least comprises: graph, report, dashboard, thermodynamic and dynamic charts.
8. A wave motion warning device for data, comprising:
the system comprises an acquisition module, a storage module and a storage module, wherein the acquisition module is used for acquiring a historical data set of a historical time period, and the historical time period comprises: a plurality of historical moments, the historical dataset comprising: historical data corresponding to a plurality of historical moments;
the analysis module is configured to analyze the historical data set by using a preset prediction model to obtain a first prediction data set of a target time period, where the preset prediction model at least includes: the method comprises the steps of presetting an autoregressive moving average model and a preset autoregressive conditional covariance model, wherein the preset autoregressive moving average model is used for predicting a second prediction data set of the target time period according to the historical data set, the preset autoregressive conditional covariance model is used for predicting an error data set of the target time period, and the target time period comprises: a plurality of target moments, the first prediction dataset comprising: at least one first prediction data corresponding to the target moment, wherein the second prediction data set comprises: at least one second prediction data corresponding to the target moment, wherein the error data set comprises: at least one error data corresponding to the target time, wherein the first prediction data is a correction result of the second prediction data and the error data of the same target time;
A monitoring module for monitoring the degree of data fluctuation of the predicted data set relative to the historical data set;
and the early warning module is used for carrying out fluctuation early warning under the condition that the fluctuation degree of the data exceeds a fluctuation threshold value.
9. A nonvolatile storage medium for storing a program, wherein the program is controlled to execute the method for warning fluctuation of data according to any one of claims 1 to 7 by a device in which the nonvolatile storage medium is located when the program is run.
10. An electronic device, comprising: a memory and a processor for executing a program stored in the processor, wherein the program executes the fluctuation warning method of data according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311801387.6A CN117744033A (en) | 2023-12-25 | 2023-12-25 | Data fluctuation early warning method and device, nonvolatile storage medium and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311801387.6A CN117744033A (en) | 2023-12-25 | 2023-12-25 | Data fluctuation early warning method and device, nonvolatile storage medium and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117744033A true CN117744033A (en) | 2024-03-22 |
Family
ID=90257480
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311801387.6A Pending CN117744033A (en) | 2023-12-25 | 2023-12-25 | Data fluctuation early warning method and device, nonvolatile storage medium and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117744033A (en) |
-
2023
- 2023-12-25 CN CN202311801387.6A patent/CN117744033A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106951984B (en) | Dynamic analysis and prediction method and device for system health degree | |
| CN110400022B (en) | Cash consumption prediction method and device for self-service teller machine | |
| EP3316198A1 (en) | Method and system for managing performance indicators for addressing goals of enterprise facility operations management | |
| CN106598791B (en) | Industrial equipment fault preventive identification method based on machine learning | |
| CN110020770A (en) | Risk and information management based on artificial intelligence | |
| US20190180379A1 (en) | Life insurance system with fully automated underwriting process for real-time underwriting and risk adjustment, and corresponding method thereof | |
| CN110738527A (en) | feature importance ranking method, device, equipment and storage medium | |
| US20210303436A1 (en) | Machine learning based data monitoring | |
| CN113570396A (en) | Time series data abnormity detection method, device, equipment and storage medium | |
| US11775412B2 (en) | Machine learning models applied to interaction data for facilitating modifications to online environments | |
| CN117196066A (en) | Intelligent operation and maintenance information analysis model | |
| CN115358155A (en) | Power big data abnormity early warning method, device, equipment and readable storage medium | |
| JP2009086896A (en) | Failure prediction system and failure prediction method for computer | |
| CN113780329A (en) | Method, apparatus, server and medium for identifying data anomalies | |
| CN117473048B (en) | Financial abnormal data monitoring and analyzing system and method based on data mining | |
| WO2020157927A1 (en) | Diagnosis system and diagnosis method | |
| CN114138601A (en) | Service alarm method, device, equipment and storage medium | |
| CN110413482B (en) | Detection method and device | |
| Terceño et al. | Prediction of business failure with fuzzy models | |
| JP7062505B2 (en) | Equipment management support system | |
| CN117744033A (en) | Data fluctuation early warning method and device, nonvolatile storage medium and electronic equipment | |
| CN113657648B (en) | Multi-dimensional data fusion equipment health assessment method, device and operation and maintenance system | |
| CN115376692A (en) | Health data processing method and device, electronic equipment and storage medium | |
| KR102499182B1 (en) | Loan regular auditing system using artificia intellicence | |
| CN113011596A (en) | Method, device and system for automatically updating model and electronic equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |