CN110516075B - Early warning report generation method and device based on machine learning and computer equipment - Google Patents

Early warning report generation method and device based on machine learning and computer equipment Download PDF

Info

Publication number
CN110516075B
CN110516075B CN201910662226.0A CN201910662226A CN110516075B CN 110516075 B CN110516075 B CN 110516075B CN 201910662226 A CN201910662226 A CN 201910662226A CN 110516075 B CN110516075 B CN 110516075B
Authority
CN
China
Prior art keywords
data
value
preset
difference
prediction
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.)
Active
Application number
CN201910662226.0A
Other languages
Chinese (zh)
Other versions
CN110516075A (en
Inventor
黄鑫
向勇
林舒
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Ping An Technology Shenzhen Co Ltd
Original Assignee
Ping An Technology Shenzhen Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Ping An Technology Shenzhen Co Ltd filed Critical Ping An Technology Shenzhen Co Ltd
Priority to CN201910662226.0A priority Critical patent/CN110516075B/en
Publication of CN110516075A publication Critical patent/CN110516075A/en
Application granted granted Critical
Publication of CN110516075B publication Critical patent/CN110516075B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
    • G06F16/367—Ontology
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9535—Search customisation based on user profiles and personalisation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Software Systems (AREA)
  • Mathematical Physics (AREA)
  • Artificial Intelligence (AREA)
  • Strategic Management (AREA)
  • Evolutionary Computation (AREA)
  • Human Resources & Organizations (AREA)
  • Computational Linguistics (AREA)
  • Computing Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Economics (AREA)
  • Animal Behavior & Ethology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Biomedical Technology (AREA)
  • Development Economics (AREA)
  • Health & Medical Sciences (AREA)
  • Game Theory and Decision Science (AREA)
  • Medical Informatics (AREA)
  • Biophysics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The application discloses an early warning report generation method, an early warning report generation device, computer equipment and a storage medium based on machine learning, wherein the method comprises the following steps: crawling initial data from an information source to obtain specified data; inputting the specified data into a preset machine learning-based trained prediction model for calculation so as to obtain a first prediction value; acquiring a second prediction numerical value by utilizing a preset knowledge graph according to the specified data; using the formula: w ═ paA+pbB, calculating a final prediction value W; calculating the difference between the final predicted value W and the predicted value of the comparison object; if the difference value is not within the preset difference value range, inputting the specified data and the data of the comparison object into a preset data difference level calculation model for calculation to obtain a data difference level value output by the data difference level calculation model, and marking the specified data of which the data difference level value is greater than a level threshold value as suspect data; and generating an early warning report.

Description

Early warning report generation method and device based on machine learning and computer equipment
Technical Field
The present application relates to the field of computers, and in particular, to a method and an apparatus for generating an early warning report based on machine learning, a computer device, and a storage medium.
Background
With the continuous development of the informatization and intelligent society, the importance of the early warning in the future is higher and higher. How to accurately predict future conditions and give early warning is a problem which needs to be solved urgently in many aspects such as finance. In the conventional technology, manual experience prediction is generally adopted, or a single prediction model is adopted for prediction and then early warning is performed, and due to the fact that data for prediction is not appropriate and the model is single, the accuracy of the result of prediction and early warning performed by the single prediction model in the conventional technology needs to be improved. And the early warning technology that traditional technology can't be based on numerical value that sets up in advance generally belongs to mechanical static early warning, can't accomplish dynamic early warning.
Disclosure of Invention
The application mainly aims to provide an early warning report generation method and device based on machine learning, computer equipment and a storage medium, and aims to improve prediction accuracy and realize dynamic early warning.
In order to achieve the above object, the present application provides a method for generating an early warning report based on machine learning, which is applied to an early warning terminal, and includes the following steps:
crawling initial data from a preset information source by adopting a preset crawler technology, and performing noise reduction processing on the initial data by using a preset noise reduction algorithm to obtain specified data, wherein the information source at least comprises a preset website;
inputting the specified data into a preset machine learning-based trained prediction model for calculation so as to obtain a first prediction value output by the prediction model; the prediction model is trained on historical data of the same type as the specified data and prediction values related to the historical data;
acquiring a second prediction numerical value output by the knowledge graph by utilizing the mutual influence relation of all knowledge nodes in a preset knowledge graph according to the specified data, wherein the knowledge graph at least comprises the knowledge nodes corresponding to the specified data;
using the formula: w ═ paA+pbB calculating a final predicted value W, where A is the first predicted value, B is the second predicted value, pa、pbWeighting parameters of the first predicted value A and the second predicted value B respectively;
calculating a difference value between the final predicted value W and a predicted value of a preset comparison object, and judging whether the difference value is within a preset difference value range, wherein the predicted value of the comparison object is obtained based on data prediction of the comparison object, and the data of the comparison object corresponds to the specified data;
if the difference value is not within a preset difference value range, inputting the specified data and the data of the comparison object into a preset data difference level calculation model for calculation so as to obtain a data difference level value output by the data difference level calculation model, and marking the specified data of which the data difference level value is greater than a preset level threshold value as suspect data;
and generating an early warning report, wherein the suspicion data is attached to the early warning report.
The application provides an early warning report generation device based on machine learning is applied to early warning terminal, includes:
the system comprises a designated data acquisition unit, a processing unit and a processing unit, wherein the designated data acquisition unit is used for crawling initial data from a preset information source by adopting a preset crawler technology and carrying out noise reduction processing on the initial data by using a preset noise reduction algorithm so as to obtain designated data, and the information source at least comprises a preset website;
a first prediction value obtaining unit, configured to input the specified data into a preset machine learning-based trained prediction model for calculation, so as to obtain a first prediction value output by the prediction model; the prediction model is trained on historical data of the same type as the specified data and prediction values related to the historical data;
a second prediction value obtaining unit, configured to obtain, according to the specified data, a second prediction value output by the knowledge graph by using a mutual influence relationship between knowledge nodes in a preset knowledge graph, where the knowledge graph at least includes the knowledge node corresponding to the specified data;
a final predicted value obtaining unit for using a formula: w ═ paA+pbB calculating a final predicted value W, where A is the first predicted value, B is the second predicted value, pa、pbWeighting parameters of the first predicted value A and the second predicted value B respectively;
a difference value calculating unit, configured to calculate a difference value between the final predicted value W and a predicted value of a preset comparison object, and determine whether the difference value is within a preset difference value range, where the predicted value of the comparison object is predicted based on data of the comparison object, and the data of the comparison object corresponds to the specified data;
a suspect data obtaining unit, configured to, if the difference is not within a preset difference range, input the specified data and the data of the comparison object into a preset data difference level calculation model for calculation, so as to obtain a data difference level value output by the data difference level calculation model, and mark the specified data, of which the data difference level value is greater than a preset level threshold, as suspect data;
and the early warning report generating unit is used for generating an early warning report, wherein the early warning report is attached with the suspicion data.
The present application provides a computer device comprising a memory storing a computer program and a processor implementing the steps of any of the above methods when the processor executes the computer program.
The present application provides a computer-readable storage medium having stored thereon a computer program which, when being executed by a processor, carries out the steps of the method of any of the above.
According to the early warning report generation method and device based on machine learning, the computer equipment and the storage medium, designated data are obtained by crawling initial data and performing noise reduction processing; inputting the specified data into a preset prediction model for calculation so as to obtain a first prediction value; acquiring a second prediction value output by the knowledge graph by utilizing the mutual influence relation of all knowledge nodes in a preset knowledge graph; using the formula: w ═ paA+pbB, calculating a final prediction value W; calculating the difference value between the final predicted value W and the predicted value of a preset comparison object; if the difference value is not within the preset difference value range, obtaining the data difference output by the data difference level calculation modelClassifying the grade value, and marking the specified data with the data difference grade value larger than a preset grade threshold value as suspect data; and generating an early warning report, wherein the suspicion data is attached to the early warning report. Therefore, the prediction accuracy is improved and the dynamic early warning is realized.
Drawings
Fig. 1 is a schematic flowchart of a method for generating a warning report based on machine learning according to an embodiment of the present application;
fig. 2 is a block diagram schematically illustrating a structure of a machine learning-based warning report generation apparatus according to an embodiment of the present application;
fig. 3 is a block diagram illustrating a structure of a computer device according to an embodiment of the present application.
The implementation, functional features and advantages of the objectives of the present application will be further explained with reference to the accompanying drawings.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
Referring to fig. 1, an embodiment of the present application provides a method for generating an early warning report based on machine learning, which is applied to an early warning terminal, and includes:
s1, crawling initial data from a preset information source by adopting a preset crawler technology, and performing noise reduction processing on the initial data by using a preset noise reduction algorithm to obtain specified data, wherein the information source at least comprises a preset website;
s2, inputting the specified data into a preset machine learning-based trained prediction model for calculation, so as to obtain a first prediction value output by the prediction model; the prediction model is trained on historical data of the same type as the specified data and prediction values related to the historical data;
s3, acquiring a second prediction numerical value output by the knowledge graph according to the designated data by utilizing the mutual influence relationship of all knowledge nodes in a preset knowledge graph, wherein the knowledge graph at least comprises the knowledge nodes corresponding to the designated data;
s4, using the formula: w ═ paA+pbB calculating a final predicted value W, where A is the first predicted value, B is the second predicted value, pa、pbWeighting parameters of the first predicted value A and the second predicted value B respectively;
s5, calculating a difference value between the final predicted value W and a preset predicted value of a comparison object, and judging whether the difference value is within a preset difference value range, wherein the predicted value of the comparison object is obtained based on data prediction of the comparison object, and the data of the comparison object and the specified data are corresponding to each other;
s6, if the difference value is not within a preset difference value range, inputting the specified data and the data of the comparison object into a preset data difference level calculation model for calculation so as to obtain a data difference level value output by the data difference level calculation model, and marking the specified data of which the data difference level value is greater than a preset level threshold value as suspect data;
and S7, generating an early warning report, wherein the suspicion data is attached to the early warning report.
As described in step S1, the initial data is crawled from a preset information source by using a preset crawler technology, and the initial data is subjected to noise reduction processing by using a preset noise reduction algorithm, so as to obtain the specified data, where the information source at least includes a preset website. The early warning terminal is used as an execution main body, wherein the early warning terminal can be any terminal and comprises a client, a server and the like. The predetermined crawler technology may be any technology, for example, a script framework in Python language is used to crawl initial data in an information source. The information source is, for example, a preset website, a software APP or a database. The initial data is data that can be a basis for prediction, such as current financial data including social security data, major project data, or current product price. The predetermined website may be any website, preferably a certified government website (i.e., a website having a digital certificate issued by a certificate authority on behalf of a government agency). And carrying out noise reduction processing on the initial data by using a preset noise reduction algorithm so as to obtain specified data. The initial data directly crawled is the first-hand data, the quantity is large and complicated, and the problems of repeated data and inaccurate data are inevitable, so that the noise reduction processing is performed on the initial data by using a preset noise reduction algorithm. The noise reduction algorithm is for example: the method comprises the steps of forming a specified value group by using values of the same initial data from different data sources, calculating the variance of each initial data in the specified value group by using a preset variance formula, judging whether the variance of each initial data in the specified value group is smaller than a preset variance threshold value, and if the variance unevenness of each initial data in the specified value group is smaller than the preset variance threshold value, taking the initial data of which the variance is not smaller than the preset variance threshold value as noise and carrying out removal processing.
Inputting the specified data into a preset machine learning-based trained prediction model for calculation, so as to obtain a first predicted value output by the prediction model, as described in the step S2 above; the prediction model is trained on historical data of the same type as the specified data and prediction values related to the historical data. Wherein the predictive model is configured to output a first predicted value for characterizing a future condition. The first predicted value is, for example, predicted financial data (budget, etc.), and further, the first predicted value is, for example, a value mapped by the predicted financial data. The prediction model is, for example, a neural network model, a support vector machine, a decision tree, etc., and the application prefers the neural network model, including: the model comprises a VGG19 model, a VGG-F model, a ResNet50 model, a DPN131 model, an inclusion V3 model, an Xception model, an AlexNet model and the like, wherein the DPN model is preferably used in the application, and the DPN (Dual Path network) is a neural network structure, and the core content of DenseNet is introduced on the basis of ResNeXt, so that the model can more fully utilize the characteristics. The DPN, resenext, and DenseNet are conventional network structures, and are not described herein. Wherein the training data includes historical data of the same type as the specified data, and predicted values associated with the historical data. Since the type of the historical data is the same as the type of the specified data (for example, the price of the same commodity and the number of the standing population in the same area), the first prediction value can be directly and accurately obtained by inputting the specified data into the prediction model.
As described in step S3, according to the specified data, a second predicted numerical value output by the knowledge graph is obtained by using the interaction relationship between the knowledge nodes in a preset knowledge graph, where the knowledge graph includes at least the knowledge node corresponding to the specified data. A knowledge graph is a series of various graphs showing the relationship between the progress and structure of knowledge development, has a plurality of knowledge nodes (e.g., a commodity, a population of a region, estimated financial data, etc.), and contains the influence relationship of the knowledge nodes. The knowledge graph establishes a relation between different knowledge nodes in an economic industry chain, and when the relation is established, the attribute of the relation and the attribute of an associated node establish a prediction connection model, wherein the model and a cost conduction model are pulled by requirements, for example, the price increase of an upstream product of an industry affects the price of a downstream product through the cost conduction model. Accordingly, according to the specified data, a second prediction numerical value output by the knowledge graph is obtained by utilizing the mutual influence relation of all knowledge nodes in the preset knowledge graph. Wherein the second predicted value is of the same type as the first predicted value (i.e. for example as a mapped value of a financial forecast expenditure).
As described above in step S4, the formula is used: w ═ paA+pbB calculating a final predicted value W, where A is the first predicted value, B is the second predicted value, pa、pbThe weight parameters are the first predicted value a and the second predicted value B respectively. In order to ensure the accuracy and stability of the predicted numerical value, the formula is used in the application: w ═ paA+pbB calculating a final predicted value W, where A is the firstPredicted value, B being said second predicted value, pa、pbThe weight parameters are the first predicted value a and the second predicted value B respectively. Therefore, the accuracy and stability of the predicted numerical value are ensured by using a dual-model weight output mode. Wherein the weighting parameters p of the first predicted value A and the second predicted value Ba、pbThe historical prediction accuracy of the knowledge graph can be set correspondingly through the corresponding prediction model, and the numerical value of the weight parameter corresponding to the higher historical prediction accuracy is larger, and the numerical value of the weight parameter corresponding to the higher historical prediction accuracy is smaller.
As described in the above step S5, a difference between the final predicted value W and a predicted value of a preset comparison object is calculated, and it is determined whether the difference is within a preset difference range, where the predicted value of the comparison object is predicted based on data of the comparison object, and the data of the comparison object and the specified data correspond to each other. Where the comparison object is equivalent to the object for which the present application predicts, e.g., where the present application predicts the economic value of one city, then the comparison object is another city, and the predicted value of the comparison object is the economic value of the other city. According to the method and the device, whether early warning is needed or not is judged by adopting the comparison result with the comparison object, so that dynamic early warning is realized, and analysis of the dynamic difference between the two objects is facilitated. The predicted value of the control object is predicted based on the data of the control object, and may be predicted by the same acquisition method as the final predicted value W. The data of the comparison object and the specified data are mutually corresponding, namely the data of the comparison object and the specified data are the same in type, for example, the specified data comprise total regional production value, common budget income, total import and export amount and the like, and the data of the comparison object also comprise the same type of data, so that the accuracy of the process between the two objects is improved.
As described in step S6, if the difference is not within the preset difference range, the data of the designated data and the data of the comparison object are input into a preset data difference level calculation model for calculation, so as to obtain the data difference level value output by the data difference level calculation model, and the designated data with the data difference level value greater than the preset level threshold value is regarded as suspect data. If the difference value is not within the preset difference value range, the difference between the prediction object and the comparison object is out of expectation, and therefore the data adopted for prediction is corresponding, and therefore suspected data can be considered to exist in the specified data of the prediction object, and the suspected data has a large influence on the condition that the difference value is not within the preset difference value range. And determining the suspect data in the specified data by inputting the specified data and the data of the comparison object into a preset data difference level calculation model for calculation so as to obtain a data difference level value output by the data difference level calculation model, and marking the specified data of which the data difference level value is greater than a preset level threshold as suspect data. The data of the designated data and the data of the comparison object are input into a preset data difference level calculation model for calculation, for example, the following modes are adopted: judging whether the specified data and the data of the comparison object belong to the same magnitude; if the specified data and the data of the comparison object belong to the same magnitude, subtracting the data of the comparison object from the specified data to obtain a data difference value, and outputting a data difference level value according to a preset mapping relation between the data difference value and the data difference level; if the specified data and the data of the comparison object do not belong to the same magnitude, according to a formula: and (3) the data difference value lg designates data-lg comparison object data, the data difference value is calculated, and a data difference grade value is output according to the preset mapping relation between the data difference value and the data difference grade.
As described in step S7, an early warning report is generated, wherein the suspect data is attached to the early warning report. Since the condition that the difference value is not within the preset difference value range is greatly influenced by the suspected data, the suspected data is attached to the early warning report. Further, in order to provide more accurate information reference, the early warning report is also attached with a historical event corresponding to the suspect data, wherein the historical event is obtained by inquiring the type of the suspect data in a preset database, and the historical event of which the type of the data corresponds to the type of the data is prestored in the database.
In one embodiment, the step S1 of using a preset crawler technology to crawl initial data from a preset information source and perform noise reduction processing on the initial data by using a preset noise reduction algorithm includes:
s101, crawling initial data in a preset website by adopting a Scapy frame in a Python language;
s102, combining the numerical values of the initial data of the same type into a specified numerical value group, and adopting a preset formula:
Figure BDA0002138926220000081
computing a global variance of the mth initial data in the specified set of values
Figure BDA0002138926220000082
Wherein N is the total number of specified values in the specified value group, Am is the value of the mth initial data, and B is the average value of the specified value group;
s103, judging the total variance
Figure BDA0002138926220000083
Whether all are smaller than a preset variance threshold value;
s104, if the total variance
Figure BDA0002138926220000084
If the unevenness is less than the preset variance threshold value, the total variance is calculated
Figure BDA0002138926220000085
And taking the initial data not less than a preset variance threshold value as noise and performing removal processing.
As described above, it is realized that the initial data is subjected to noise reduction processing using a preset noise reduction algorithm, thereby obtaining the specified data. The script framework of the Python language is an effective means for crawling information in a preset website, and mainly comprises the following steps: guiding deviceAn engine, a scheduler, a downloader, a crawler, a project pipeline, downloader middleware, crawler middleware, scheduling middleware, etc. The specific crawling process comprises the following steps: the engine fetches a link from the scheduler for the next fetch; the engine encapsulates the link into a request and transmits the request to the downloader; downloading the resource by the downloader; the crawler analyzes the entity and gives the entity to the entity pipeline for further processing. Because inaccurate data may exist in the crawled numerical value, the method adopts a preset formula:
Figure BDA0002138926220000086
computing a global variance of the mth initial data in the specified set of values
Figure BDA0002138926220000087
Judging the total variance
Figure BDA0002138926220000088
Whether all are smaller than a preset variance threshold value; if the total variance
Figure BDA0002138926220000089
If the unevenness is less than the preset variance threshold value, the total variance is calculated
Figure BDA0002138926220000091
And taking the initial data not less than a preset variance threshold value as noise and performing removal processing. Therefore, the problem of prediction misalignment caused by inaccurate data is avoided.
In one embodiment, the specified data is input into a preset machine learning-based trained prediction model for calculation, so as to obtain a first prediction value output by the prediction model; before step S2, in which the prediction model is trained based on the historical data of the same type as the specified data and the prediction value associated with the historical data, the method includes:
s11, obtaining sample data with specified volume, and dividing the sample data into a training set and a test set; wherein the sample data comprises historical data of the same type as the specified data, and a predicted numerical value associated with the historical data;
s12, inputting the sample data of the training set into a preset neural network model for training; wherein, a random gradient descent method is adopted in the training process to obtain an initial training model;
s13, verifying the initial training model by using the sample data of the test set;
and S14, if the verification is passed, marking the initial training model as the prediction model.
As described above, training the predictive model is implemented. The present application adopts a neural network model as a machine learning model, including, for example: VGG19 model, VGG-F model, ResNet50 model, DPN131 model, inclusion v3 model, Xception model, AlexNet model, etc., and the DPN model is more preferably used in the present application. Before training, sample data is randomly divided into two sets, namely a training set and a testing set, and then the neural network model is trained through the sample data of the training set. And after the training is finished to obtain a result training model, verifying the result training model through sample data of the test set to judge whether the result training model is available. The random gradient descent method is to randomly sample some training data to replace the whole training set, so as to increase the training speed. Further, parameters of each layer of the neural network model can be updated by using a back propagation rule (BP), which is based on a gradient descent method, and an input-output relationship of the BP network is substantially a mapping relationship: the function of a BP neural network with n input and m output is continuous mapping from n dimension Euclidean space to a finite field in m dimension Euclidean space, and the mapping has high nonlinearity, thereby realizing the updating of parameters of each layer of a neural network model.
In one embodiment, the use formula: w ═ paA+pbB calculating a final predicted value W, where A is the first predicted value, B is the second predicted value, pa、pbStep S4 of weighting parameters of the first predicted value A and the second predicted value B respectivelyPreviously, comprising:
s31, acquiring a first historical prediction numerical value output by the prediction model, a second historical prediction numerical value output by the knowledge graph and a historical real numerical value;
s32, according to the formula: calculating a first prediction deviation value from a first historical prediction value to a historical true value;
s33, according to the formula: calculating a second prediction deviation value from a second historical prediction value to a historical real value;
s34, according to the first prediction deviation value and the second prediction deviation value, respectively obtaining the weight parameters p corresponding to the first prediction value A and the second prediction value B by using the corresponding relation between the preset prediction deviation value and the weight parametersa、pb。
As described above, obtaining the weight parameter p corresponding to the first predicted numerical value a and the second predicted numerical value B is realizeda、pb. The more accurate the output value of the prediction model is, the higher the corresponding weight parameter is, so that the accuracy of the final predicted value W is higher. The application is characterized in that the formula is as follows: the first predicted deviation value is | the first historical predicted value-the historical true value |, by the equation: respectively obtaining the weight parameters p corresponding to the first predicted value a and the second predicted value B by using a preset corresponding relationship between the predicted deviation value and the weight parameter in a manner of | the second predicted deviation value-the second historical true value |a、pb. Wherein the historical true values are true accurate values corresponding to the first historical predicted values and the second historical predicted values for determining the prediction accuracy of the prediction model and the knowledge graph.
In one embodiment, the step S5 of calculating a difference between the final predicted value W and a predicted value of a preset comparison object predicted based on data of the comparison object corresponding to the designated data includes:
s51, if the difference value is within a preset difference value range, judging whether the final predicted value W is within a preset value range;
s52, if the final predicted value W is not in a preset value range, acquiring the first predicted value and/or the second predicted value which are not in the preset value range, and recording the first predicted value and/or the second predicted value as a special value;
and S53, generating an early warning report, wherein the early warning report is attached with the final prediction value W, the first prediction value and the second prediction value, and the special value is specially marked in the early warning report.
As described above, generation of an early warning report and special marking of the special numerical value in the early warning report is achieved. If the difference value is within the preset difference value range, the difference value is not greatly different from the comparison object. In order to further warn, the method for judging whether the final prediction value W is within a preset value range is adopted. If the final predicted value W is not within the predetermined range, it may be that more than one of the prediction model and the knowledge-graph has a large prediction deviation. Therefore, the first prediction value and/or the second prediction value which are not within the preset value range are/is obtained and recorded as special values, and then an early warning report is generated, wherein the early warning report is attached with the final prediction value, the first prediction value and the second prediction value, and the special values are specially marked in the early warning report. Therefore, the special value of the special mark can be regarded as important, and processing measures such as rechecking can be carried out. Further, the method comprises the step of rechecking the special value of the special mark, and if the special value is confirmed to be correct, the predicted future condition is severe and early preparation is needed.
In one embodiment, the step S6, if the difference is not within a preset difference range, of inputting the data of the designated data and the data of the comparison object into a preset data difference level calculation model for calculation to obtain a data difference level value output by the data difference level calculation model, and marking the designated data of which the data difference level value is greater than a preset level threshold as suspect data, includes:
s601, if the difference value is not within a preset difference value range, judging whether the specified data and the data of the comparison object belong to the same magnitude;
s602, if the specified data and the data of the comparison object belong to the same magnitude, subtracting the data of the comparison object from the specified data to obtain a first data difference value, and outputting a data difference level value according to a preset mapping relation between the first data difference value and a data difference level;
s603, if the specified data and the data of the comparison object do not belong to the same magnitude, according to a formula: the second data difference value lg designates data-lg comparison object data, calculates a second data difference value, and outputs a data difference level value according to a preset mapping relation between the second data difference value and the data difference level;
s604, marking the specified data with the data difference level value larger than the preset level threshold value as suspect data.
As described above, it is achieved that designated data having a data difference level value larger than a preset level threshold value is marked as suspect data. The method comprises the steps of dividing the designated data into two types, wherein one type of data belongs to the same magnitude as that of the data of the comparison object, so that subtraction processing can be directly adopted, and then a data difference grade value is output according to a preset mapping relation between a first data difference value and a data difference grade; and the other class of data is not in the same magnitude as the data of the control object, according to the formula: and the second data difference is lg designated data-lg comparison target data, the second data difference is calculated, and a data difference level value is output according to a preset mapping relation between the second data difference and the data difference level. Thereby improving the accuracy of the difference judgment between the specified data and the data of the comparison object. Where lg is a log function of base 10.
In one embodiment, the generating the warning report by the warning terminal being a blockchain node in a pre-constructed blockchain network, where after the step S7 of attaching the suspect data to the warning report, the generating the warning report includes:
s71, sending the pre-warning report to an audit block chain node in the block chain network, and requiring the audit block chain node to audit the pre-warning report;
s72, receiving an audit result returned by the audit block chain node, and judging whether the audit result meets a preset block chain recording condition;
and S73, if the auditing result meets the preset block chain recording condition, recording the early warning report into the block chain network.
As described above, logging of the early warning report into a public ledger in the blockchain network is achieved. Wherein the pre-constructed blockchain network may be a public chain, a federation chain, or a private chain. The preset consensus mechanism of the block chain network may be any consensus mechanism, such as a workload certification mechanism, a rights and interests certification mechanism, a byzantine fault tolerance mechanism, a share authorization certification mechanism, and the like. If the audit result meets the preset condition of block chain record, most block chain nodes can recognize the early warning report, so that the early warning report can be recorded in a public account book in the block chain network. The preset blockchain recording conditions are, for example: and returning an audit result that the total number of the audit block chain nodes passing the audit is greater than a preset audit number threshold. And recording the early warning report into a public account book in the block chain network, so that the safety, authority and easy-to-call technical effects of the early warning report are ensured by means of the characteristics of difficult tampering and distributed storage of the block chain network.
According to the early warning report generation method based on machine learning, designated data are obtained by crawling initial data and performing noise reduction processing; inputting the specified data into a preset prediction modelCalculating to obtain a first predicted value; acquiring a second prediction value output by the knowledge graph by utilizing the mutual influence relation of all knowledge nodes in a preset knowledge graph; using the formula: w ═ paA+pbB, calculating a final prediction value W; calculating the difference value between the final predicted value W and the predicted value of a preset comparison object; if the difference value is not within a preset difference value range, obtaining a data difference level value output by the data difference level calculation model, and marking the specified data of which the data difference level value is greater than a preset level threshold value as suspect data; and generating an early warning report, wherein the suspicion data is attached to the early warning report. Therefore, the prediction accuracy is improved and the dynamic early warning is realized.
Referring to fig. 2, an embodiment of the present application provides an early warning report generation apparatus based on machine learning, which is applied to an early warning terminal, and includes:
the system comprises a specified data acquisition unit 10, a processing unit and a processing unit, wherein the specified data acquisition unit is used for crawling initial data from a preset information source by adopting a preset crawler technology, and performing noise reduction processing on the initial data by using a preset noise reduction algorithm so as to obtain specified data, and the information source at least comprises a preset website;
a first prediction value obtaining unit 20, configured to input the specified data into a preset machine learning-based trained prediction model for calculation, so as to obtain a first prediction value output by the prediction model; the prediction model is trained on historical data of the same type as the specified data and prediction values related to the historical data;
a second predicted numerical value obtaining unit 30, configured to obtain, according to the specified data, a second predicted numerical value output by the knowledge graph by using an interaction relationship between knowledge nodes in a preset knowledge graph, where the knowledge graph at least includes the knowledge node corresponding to the specified data;
a final predicted value obtaining unit 40 for obtaining a final predicted value using the formula: w ═ paA+pbB calculating a final predicted value W, where A is the first predicted value and B is the second predicted valueTwo predicted values, pa、pbWeighting parameters of the first predicted value A and the second predicted value B respectively;
a difference value calculating unit 50, configured to calculate a difference value between the final predicted value W and a predicted value of a preset comparison object, and determine whether the difference value is within a preset difference value range, where the predicted value of the comparison object is predicted based on data of the comparison object, and the data of the comparison object and the specified data correspond to each other;
a suspected data obtaining unit 60, configured to, if the difference is not within a preset difference range, input the specified data and the data of the comparison object into a preset data difference level calculation model for calculation, so as to obtain a data difference level value output by the data difference level calculation model, and mark the specified data, of which the data difference level value is greater than a preset level threshold, as suspected data;
an early warning report generating unit 70, configured to generate an early warning report, where the early warning report is attached with the suspicion data.
The operations performed by the units are respectively in one-to-one correspondence with the steps of the method for generating an early warning report based on machine learning according to the foregoing embodiment, and are not described herein again.
In one embodiment, the specified data acquiring unit 10 includes:
the crawling initial data subunit is used for crawling initial data in a preset website by adopting a Scapy framework of a Python language;
the overall variance calculating subunit is used for combining the values of the initial data of the same type into a specified value group and adopting a preset formula:
Figure BDA0002138926220000141
computing a global variance of the mth initial data in the specified set of values
Figure BDA0002138926220000142
Where N is the total number of specified values in the specified set of values and Am isThe value of the mth initial data, B being the average of the set of specified values;
a variance threshold judgment subunit for judging the total variance
Figure BDA0002138926220000143
Whether all are smaller than a preset variance threshold value;
a removal processing subunit for determining the overall variance
Figure BDA0002138926220000144
If the unevenness is less than the preset variance threshold value, the total variance is calculated
Figure BDA0002138926220000145
And taking the initial data not less than a preset variance threshold value as noise and performing removal processing.
The operations respectively executed by the sub-units correspond to the steps of the method for generating the warning report based on machine learning in the foregoing embodiment one by one, and are not described herein again.
In one embodiment, the apparatus comprises:
the system comprises a sample data acquisition unit, a training set acquisition unit and a test set acquisition unit, wherein the sample data acquisition unit is used for acquiring sample data with a specified amount and dividing the sample data into a training set and a test set; wherein the sample data comprises historical data of the same type as the specified data, and a predicted numerical value associated with the historical data;
the training unit is used for inputting the sample data of the training set into a preset neural network model for training; wherein, a random gradient descent method is adopted in the training process to obtain an initial training model;
a verification unit for verifying the initial training model by using the sample data of the test set;
and the marking unit is used for marking the initial training model as the prediction model if the verification is passed.
The operations performed by the units are respectively in one-to-one correspondence with the steps of the method for generating an early warning report based on machine learning according to the foregoing embodiment, and are not described herein again.
In one embodiment, the apparatus comprises:
the numerical value acquisition unit is used for acquiring a first historical prediction numerical value output by the prediction model, a second historical prediction numerical value output by the knowledge graph and a historical real numerical value;
a first predicted deviation value calculation unit for calculating a first predicted deviation value by the formula: calculating a first prediction deviation value from a first historical prediction value to a historical true value;
a second predicted deviation value calculation unit for calculating a second predicted deviation value by the formula: calculating a second prediction deviation value from a second historical prediction value to a historical real value;
a weight parameter obtaining unit, configured to obtain, according to the first predicted deviation value and the second predicted deviation value, a weight parameter p corresponding to the first predicted value a and the second predicted value B respectively by using a preset corresponding relationship between the predicted deviation value and the weight parametera、pb。
The operations performed by the units are respectively in one-to-one correspondence with the steps of the method for generating an early warning report based on machine learning according to the foregoing embodiment, and are not described herein again.
In one embodiment, the apparatus comprises:
a preset value range determination unit, configured to determine whether the final predicted value W is within a preset value range if the difference is within the preset difference range;
a special value marking unit, configured to, if the final predicted value W is not within a preset value range, obtain the first predicted value and/or the second predicted value that are not within the preset value range, and mark the first predicted value and/or the second predicted value as a special value;
and the early warning report generating unit is used for generating an early warning report, wherein the early warning report is attached with the final prediction value W, the first prediction value and the second prediction value, and the special value is specially marked in the early warning report.
The operations performed by the units are respectively in one-to-one correspondence with the steps of the method for generating an early warning report based on machine learning according to the foregoing embodiment, and are not described herein again.
In one embodiment, the suspected data obtaining unit 60 includes:
the same-magnitude judging subunit is configured to judge whether the specified data and the data of the comparison object belong to the same magnitude if the difference is not within a preset difference range;
a first data difference level value output subunit, configured to subtract the data of the comparison object from the designated data to obtain a first data difference value if the designated data and the data of the comparison object are of the same magnitude, and output a data difference level value according to a preset mapping relationship between the first data difference value and the data difference level;
a second data difference level value output subunit, configured to, if the specified data and the data of the comparison object do not belong to the same level, according to a formula: the second data difference value lg designates data-lg comparison object data, calculates a second data difference value, and outputs a data difference level value according to a preset mapping relation between the second data difference value and the data difference level;
and the suspected data marking subunit is used for marking the specified data with the data difference level value larger than a preset level threshold value as suspected data.
The operations respectively executed by the sub-units correspond to the steps of the method for generating the warning report based on machine learning in the foregoing embodiment one by one, and are not described herein again.
In one embodiment, the apparatus comprises:
the early warning report sending unit is used for sending the early warning report to an audit block chain link point in the block chain network and requesting the audit block chain link point to audit the early warning report;
the audit result receiving unit is used for receiving the audit result returned by the audit block chain node and judging whether the audit result meets the preset block chain recording condition;
and the audit result recording unit is used for recording the early warning report into the block chain network if the audit result meets the preset block chain recording condition.
The operations performed by the units are respectively in one-to-one correspondence with the steps of the method for generating an early warning report based on machine learning according to the foregoing embodiment, and are not described herein again.
The early warning report generation device based on machine learning obtains specified data by crawling initial data and performing noise reduction processing; inputting the specified data into a preset prediction model for calculation so as to obtain a first prediction value; acquiring a second prediction value output by the knowledge graph by utilizing the mutual influence relation of all knowledge nodes in a preset knowledge graph; using the formula: w ═ paA+pbB, calculating a final prediction value W; calculating the difference value between the final predicted value W and the predicted value of a preset comparison object; if the difference value is not within a preset difference value range, obtaining a data difference level value output by the data difference level calculation model, and marking the specified data of which the data difference level value is greater than a preset level threshold value as suspect data; and generating an early warning report, wherein the suspicion data is attached to the early warning report. Therefore, the prediction accuracy is improved and the dynamic early warning is realized.
Referring to fig. 3, an embodiment of the present invention further provides a computer device, where the computer device may be a server, and an internal structure of the computer device may be as shown in the figure. The computer device includes a processor, a memory, a network interface, and a database connected by a system bus. Wherein the computer designed processor is used to provide computational and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used for storing data used by the early warning report generation method based on machine learning. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement a machine learning-based early warning report generation method.
The processor executes the method for generating the early warning report based on machine learning, and is applied to the early warning terminal, wherein the steps included in the method correspond to the steps of executing the method for generating the early warning report based on machine learning in the foregoing embodiment one to one, and are not described herein again.
The computer equipment acquires the specified data by crawling the initial data and performing noise reduction processing; inputting the specified data into a preset prediction model for calculation so as to obtain a first prediction value; acquiring a second prediction value output by the knowledge graph by utilizing the mutual influence relation of all knowledge nodes in a preset knowledge graph; using the formula: w ═ paA+pbB, calculating a final prediction value W; calculating the difference value between the final predicted value W and the predicted value of a preset comparison object; if the difference value is not within a preset difference value range, obtaining a data difference level value output by the data difference level calculation model, and marking the specified data of which the data difference level value is greater than a preset level threshold value as suspect data; and generating an early warning report, wherein the suspicion data is attached to the early warning report. Therefore, the prediction accuracy is improved and the dynamic early warning is realized.
An embodiment of the present application further provides a computer-readable storage medium, where a computer program is stored thereon, and when the computer program is executed by a processor, the method for generating an early warning report based on machine learning is implemented, and is applied to an early warning terminal, where steps included in the method correspond to steps of executing the method for generating an early warning report based on machine learning in the foregoing embodiment one to one, and are not described herein again.
The computer-readable storage medium acquires the specified data by crawling the initial data and performing noise reduction processing; inputting the specified data into a preset prediction model for calculation so as to obtain a first prediction value; using preset knowledge-mapsObtaining a second prediction value output by the knowledge graph according to the mutual influence relationship of all knowledge nodes in the knowledge graph; using the formula: w ═ paA+pbB, calculating a final prediction value W; calculating the difference value between the final predicted value W and the predicted value of a preset comparison object; if the difference value is not within a preset difference value range, obtaining a data difference level value output by the data difference level calculation model, and marking the specified data of which the data difference level value is greater than a preset level threshold value as suspect data; and generating an early warning report, wherein the suspicion data is attached to the early warning report. Therefore, the prediction accuracy is improved and the dynamic early warning is realized.
The above description is only a preferred embodiment of the present application, and not intended to limit the scope of the present application, and all modifications of equivalent structures and equivalent processes, which are made by the contents of the specification and the drawings of the present application, or which are directly or indirectly applied to other related technical fields, are also included in the scope of the present application.

Claims (10)

1. A machine learning-based early warning report generation method is applied to an early warning terminal and is characterized by comprising the following steps:
crawling initial data from a preset information source by adopting a preset crawler technology, and performing noise reduction processing on the initial data by using a preset noise reduction algorithm to obtain specified data, wherein the information source at least comprises a preset website;
inputting the specified data into a preset machine learning-based trained prediction model for calculation so as to obtain a first prediction value output by the prediction model; the prediction model is trained on historical data of the same type as the specified data and prediction values related to the historical data;
acquiring a second prediction numerical value output by the knowledge graph by utilizing the mutual influence relation of all knowledge nodes in a preset knowledge graph according to the specified data, wherein the knowledge graph at least comprises the knowledge nodes corresponding to the specified data;
using the formula: w ═ paA+pbB calculating a final predicted value W, where A is the first predicted value, B is the second predicted value, pa、pbWeighting parameters of the first predicted value A and the second predicted value B respectively;
calculating a difference value between the final predicted value W and a predicted value of a preset comparison object, and judging whether the difference value is within a preset difference value range, wherein the predicted value of the comparison object is obtained based on data prediction of the comparison object, and the data of the comparison object corresponds to the specified data;
if the difference value is not within a preset difference value range, inputting the specified data and the data of the comparison object into a preset data difference level calculation model for calculation so as to obtain a data difference level value output by the data difference level calculation model, and marking the specified data of which the data difference level value is greater than a preset level threshold value as suspect data;
and generating an early warning report, wherein the suspicion data is attached to the early warning report.
2. The method for generating an early warning report based on machine learning according to claim 1, wherein the step of crawling initial data from a preset information source by using a preset crawler technology and performing noise reduction processing on the initial data by using a preset noise reduction algorithm comprises:
crawling initial data in a preset website by adopting a Scapy frame of Python language;
and (3) combining the values of the initial data of the same type into a specified value group, and adopting a preset formula:
Figure FDA0002138926210000021
computing a global variance of the mth initial data in the specified set of values
Figure FDA0002138926210000022
Wherein N is the specified set of valuesAm is the value of the mth initial data, and B is the average value of the specified value group;
judging the total variance
Figure FDA0002138926210000023
Whether all are smaller than a preset variance threshold value;
if the total variance
Figure FDA0002138926210000024
If the unevenness is less than the preset variance threshold value, the total variance is calculated
Figure FDA0002138926210000025
And taking the initial data not less than a preset variance threshold value as noise and performing removal processing.
3. The method for generating a machine learning-based warning report according to claim 1, wherein the specified data is input into a preset machine learning-based trained prediction model for calculation, so as to obtain a first prediction value output by the prediction model; wherein, before the step of training the prediction model based on the historical data with the same type as the specified data and the prediction value associated with the historical data, the method comprises the following steps:
obtaining sample data of a specified amount, and dividing the sample data into a training set and a test set; wherein the sample data comprises historical data of the same type as the specified data, and a predicted numerical value associated with the historical data;
inputting sample data of a training set into a preset neural network model for training; wherein, a random gradient descent method is adopted in the training process to obtain an initial training model;
verifying the initial training model by using the sample data of the test set;
and if the verification is passed, recording the initial training model as the prediction model.
4. The machine-learning based early warning report generation method of claim 1, wherein the usage formula: w ═ paA+pbB calculating a final predicted value W, where A is the first predicted value, B is the second predicted value, pa、pbBefore the step of weighting the first predicted value a and the second predicted value B, respectively, the method includes:
acquiring a first historical prediction numerical value output by the prediction model, a second historical prediction numerical value output by the knowledge graph and a historical real numerical value;
by the formula: calculating a first prediction deviation value from a first historical prediction value to a historical true value;
by the formula: calculating a second prediction deviation value from a second historical prediction value to a historical real value;
according to the first prediction deviation value and the second prediction deviation value, respectively obtaining a weight parameter p corresponding to the first prediction value A and the second prediction value B by utilizing a preset corresponding relation between the prediction deviation value and the weight parametera、pb。
5. The method for generating an early warning report based on machine learning according to claim 1, wherein the step of calculating a difference between the final predicted value W and a predicted value of a preset comparison object, and determining whether the difference is within a preset difference range, wherein the predicted value of the comparison object is predicted based on data of the comparison object, and the step of mutually corresponding the data of the comparison object and the specified data comprises:
if the difference value is within a preset difference value range, judging whether the final predicted value W is within a preset value range;
if the final predicted numerical value W is not in a preset numerical range, acquiring the first predicted numerical value and/or the second predicted numerical value which are not in the preset numerical range, and recording the first predicted numerical value and/or the second predicted numerical value as a special numerical value;
generating an early warning report, wherein the early warning report is attached with the final predicted value W, the first predicted value and the second predicted value, and the special value is specially marked in the early warning report.
6. The method for generating an early warning report based on machine learning according to claim 1, wherein if the difference is not within a preset difference range, the step of inputting the specified data and the data of the comparison object into a preset data difference level calculation model for calculation so as to obtain a data difference level value output by the data difference level calculation model, and marking the specified data with the data difference level value larger than a preset level threshold as suspect data comprises:
if the difference value is not within a preset difference value range, judging whether the specified data and the data of the comparison object belong to the same magnitude;
if the specified data and the data of the comparison object belong to the same magnitude, subtracting the data of the comparison object from the specified data to obtain a first data difference value, and outputting a data difference level value according to a preset mapping relation between the first data difference value and the data difference level;
if the specified data and the data of the comparison object do not belong to the same magnitude, according to a formula: the second data difference value lg designates data-lg comparison object data, calculates a second data difference value, and outputs a data difference level value according to a preset mapping relation between the second data difference value and the data difference level;
and marking the specified data with the data difference level value larger than a preset level threshold value as suspect data.
7. The method for generating an early warning report based on machine learning according to claim 1, wherein the early warning terminal is a blockchain node in a previously constructed blockchain network, and the generating of the early warning report, wherein the step of attaching the suspect data to the early warning report comprises:
sending the pre-warning report to an audit block chain link point in the block chain network, and requiring the audit block chain link point to audit the pre-warning report;
receiving an audit result returned by the audit block chain node, and judging whether the audit result meets a preset block chain recording condition;
and if the audit result meets the preset block chain recording condition, recording the early warning report into the block chain network.
8. The utility model provides an early warning report generation device based on machine learning, is applied to early warning terminal, its characterized in that includes:
the system comprises a designated data acquisition unit, a processing unit and a processing unit, wherein the designated data acquisition unit is used for crawling initial data from a preset information source by adopting a preset crawler technology and carrying out noise reduction processing on the initial data by using a preset noise reduction algorithm so as to obtain designated data, and the information source at least comprises a preset website;
a first prediction value obtaining unit, configured to input the specified data into a preset machine learning-based trained prediction model for calculation, so as to obtain a first prediction value output by the prediction model; the prediction model is trained on historical data of the same type as the specified data and prediction values related to the historical data;
a second prediction value obtaining unit, configured to obtain, according to the specified data, a second prediction value output by the knowledge graph by using a mutual influence relationship between knowledge nodes in a preset knowledge graph, where the knowledge graph at least includes the knowledge node corresponding to the specified data;
a final predicted value obtaining unit for using a formula: w ═ paA+pbB calculating a final predicted value W, where A is the first predicted value, B is the second predicted value, pa、pbRespectively being said first predicted value A, saidA weight parameter for the second predicted value B;
a difference value calculating unit, configured to calculate a difference value between the final predicted value W and a predicted value of a preset comparison object, and determine whether the difference value is within a preset difference value range, where the predicted value of the comparison object is predicted based on data of the comparison object, and the data of the comparison object corresponds to the specified data;
a suspect data obtaining unit, configured to, if the difference is not within a preset difference range, input the specified data and the data of the comparison object into a preset data difference level calculation model for calculation, so as to obtain a data difference level value output by the data difference level calculation model, and mark the specified data, of which the data difference level value is greater than a preset level threshold, as suspect data;
and the early warning report generating unit is used for generating an early warning report, wherein the early warning report is attached with the suspicion data.
9. A computer device comprising a memory and a processor, the memory storing a computer program, wherein the processor implements the steps of the method of any one of claims 1 to 7 when executing the computer program.
10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 7.
CN201910662226.0A 2019-07-22 2019-07-22 Early warning report generation method and device based on machine learning and computer equipment Active CN110516075B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910662226.0A CN110516075B (en) 2019-07-22 2019-07-22 Early warning report generation method and device based on machine learning and computer equipment

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910662226.0A CN110516075B (en) 2019-07-22 2019-07-22 Early warning report generation method and device based on machine learning and computer equipment

Publications (2)

Publication Number Publication Date
CN110516075A CN110516075A (en) 2019-11-29
CN110516075B true CN110516075B (en) 2022-04-15

Family

ID=68623273

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910662226.0A Active CN110516075B (en) 2019-07-22 2019-07-22 Early warning report generation method and device based on machine learning and computer equipment

Country Status (1)

Country Link
CN (1) CN110516075B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111859238A (en) * 2020-07-27 2020-10-30 平安科技（深圳）有限公司 Method and device for predicting data change frequency based on model and computer equipment

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109064313A (en) * 2018-07-20 2018-12-21 重庆富民银行股份有限公司 Warning monitoring system after the loan of knowledge based graphical spectrum technology
CN109639670A (en) * 2018-12-10 2019-04-16 北京威努特技术有限公司 A kind of industry control network security postures quantitative estimation method of knowledge based map
CN110008349A (en) * 2019-02-01 2019-07-12 阿里巴巴集团控股有限公司 The method and device for the event risk assessment that computer executes

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10839296B2 (en) * 2016-11-30 2020-11-17 Accenture Global Solutions Limited Automatic prediction of an event using data

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109064313A (en) * 2018-07-20 2018-12-21 重庆富民银行股份有限公司 Warning monitoring system after the loan of knowledge based graphical spectrum technology
CN109639670A (en) * 2018-12-10 2019-04-16 北京威努特技术有限公司 A kind of industry control network security postures quantitative estimation method of knowledge based map
CN110008349A (en) * 2019-02-01 2019-07-12 阿里巴巴集团控股有限公司 The method and device for the event risk assessment that computer executes

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
基于神经网络的知识推理研究综述;张仲伟等;《计算机工程与应用》;20190325;第8-19页 *

Also Published As

Publication number Publication date
CN110516075A (en) 2019-11-29

Similar Documents

Publication Publication Date Title
Zeng et al. A self‐adaptive intelligence gray prediction model with the optimal fractional order accumulating operator and its application
US20230195429A1 (en) Graph outcome determination in domain-specific execution environment
US11132403B2 (en) Graph-manipulation based domain-specific execution environment
CN103412918B (en) A kind of service trust degree appraisal procedure based on service quality and reputation
US11640348B2 (en) Generating anomaly alerts for time series data
Witt et al. Decentral and incentivized federated learning frameworks: A systematic literature review
Melo et al. Using blockchains to implement distributed measuring systems
US20080215576A1 (en) Fusion and visualization for multiple anomaly detection systems
Zou et al. QoS-aware web service recommendation with reinforced collaborative filtering
Malhotra et al. Prediction & assessment of change prone classes using statistical & machine learning techniques
WO2022012136A1 (en) Block chain system, and biological product monitoring method and device
CN113269359B (en) User financial status prediction method, device, medium, and computer program product
Chen et al. Your neighbors are misunderstood: On modeling accurate similarity driven by data range to collaborative web service QoS prediction
CN110516075B (en) Early warning report generation method and device based on machine learning and computer equipment
Ramakrishnan et al. An empirical comparison of predictive models for web page performance
Hole et al. Anomaly detection with htm
CN105184170B (en) A kind of domain software credibility assessment method based on formalization degree
CN116629612A (en) Risk prediction method and device, storage medium and electronic equipment
Zhang et al. Parametric sensitivity analysis for importance measure on failure probability and its efficient Kriging solution
US20220129318A1 (en) Quota Request Resolution On Computing Platform
CA3179224A1 (en) Bias detection and reduction in a machine-learning techniques
US11449409B2 (en) Schema inference and log data validation system
Su et al. Quantitative verification for monitoring event-streaming systems
Ghosheh et al. A comparative analysis of maintainability approaches for web applications
Pal et al. Blockchain network: Performance optimization

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
GR01 Patent grant
GR01 Patent grant