CN107357764A - Data analysing method, electronic equipment and computer-readable storage medium - Google Patents
Data analysing method, electronic equipment and computer-readable storage medium Download PDFInfo
- Publication number
- CN107357764A CN107357764A CN201710488327.1A CN201710488327A CN107357764A CN 107357764 A CN107357764 A CN 107357764A CN 201710488327 A CN201710488327 A CN 201710488327A CN 107357764 A CN107357764 A CN 107357764A
- Authority
- CN
- China
- Prior art keywords
- historical data
- prediction
- period
- change direction
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- 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
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
-
- 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
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
- G06Q30/0203—Market surveys; Market polls
Abstract
The embodiment of the invention discloses a kind of data analysing method, electronic equipment and computer-readable storage medium, methods described includes:Obtain the historical data of the first parameter in the first time period of history;Based on the historical data, the prediction change direction in second time period is determined, wherein, the second time period is later than the first time period;With reference to the prediction change direction and the historical data, the first parameter is calculated in the second time period, the predicted value with the prediction change direction.
Description
Technical field
The present invention relates to areas of information technology, more particularly to a kind of data analysing method, electronic equipment and computer storage
Medium.
Background technology
In information process, it may be related to and data processing is carried out based on historical data, obtain following a period of time
Interior predicted value.
In the data analysing method of existing predicted value, it will usually directly carried out using various mathematical modelings.It is for example, sharp
Historical data is analyzed with model is looked back, obtains following a cycle or the predicted value in multiple cycles.
But practice have shown that, on the one hand, the data analysing method of this predicted value, to the exceptional value in historical data very
Sensitivity, the problem of accuracy caused by easily there is over-fitting is low.On the other hand, directly prediction data is disposably located
Reason, obtained predicted value is excessively coarse, and accuracy and accuracy are all very low.
The content of the invention
In view of this, the embodiment of the present invention it is expected that providing a kind of data analysing method, electronic equipment and computer storage is situated between
Matter, at least partly solve the problems, such as that accuracy and accuracy are too low.
To reach above-mentioned purpose, the technical proposal of the invention is realized in this way:
The embodiment of the present invention provides a kind of data analysing method, including:
Obtain the historical data of the first parameter in the first time period of history;
Based on the historical data, the prediction change direction in second time period is determined, wherein, the second time period evening
In the first time period;
With reference to the prediction change direction and the historical data, the first parameter is calculated in the second time period,
Predicted value with the prediction change direction.
Alternatively, it is described that change direction and the historical data are predicted with reference to described in, the first parameter is calculated described the
In two periods, there is the predicted value of the prediction change direction, including:
With reference to the prediction change direction and the historical data, determine first parameter in the prediction change direction
On prediction change to attributes, wherein, it is described prediction change to attributes include:Predict rate of change, prediction constant interval and prediction change
At least one of rate grade;
According to the prediction change to attributes and with reference to the historical data, the first parameter is calculated in the second time period
It is interior, there is the predicted value of the prediction change to attributes.
Alternatively, it is described that change direction and the historical data are predicted with reference to described in, determine first parameter described
The prediction change to attributes in change direction is predicted, including:
With the first historical data in the historical data and the actual change direction of (n+1)th sub- period and actual become
Change attribute to be trained for training sample, so as to obtain attribute forecast model, wherein, first historical data includes:It is described
The 1st sub- period is to the historical data in n-th of sub- period in first time period;
Using the second historical data in the historical data by input data and to predict change direction, the category is input to
Property forecast model, the prediction change to attributes of the second time period is obtained, wherein, second historical data includes:2nd
The individual sub- period, n was the integer not less than 2 to the historical data in (n+1)th sub- period.
Alternatively, it is described according to the prediction change to attributes and with reference to the historical data, the first parameter is calculated in institute
State in second time period, there is the predicted value of the prediction change to attributes, including:
With first historical data, actual change direction, actual change attribute and the institute of (n+1)th sub- period
The actual value for stating the first parameter carries out model training, so as to obtain value prediction model;
With second historical data, the prediction change direction, the prediction change to attributes are as input parameter, input
To the value prediction model, the predicted value of the second time period is obtained.
Alternatively, it is described to be based on the historical data, the prediction change direction in second time period is determined, including:
Using the first historical data in the historical data and the prediction change direction of (n+1)th period as training sample
It is trained, so as to obtain direction prediction model, wherein, first historical data includes:The 1st in the first time period
The sub- period is to the historical data in n-th of sub- period;
Using the second historical data in the historical data as input data, the directional prediction modes are input to, are obtained
Second time end the prediction change direction;Wherein, second historical data includes:2nd sub- period is to n-th
The historical data in+1 sub- period, n are the integer not less than 2.
Alternatively, it is described to be based on the historical data, the prediction change direction in second time period is determined, including:
The initial value of first parameter of initial time based on the first time period, and the first time period
The stop value of second parameter of end time;
Based on the initial value and the stop value, determine that the overall of the second time period predicts change direction;
It is described that change direction and the historical data are predicted with reference to described in, determine first parameter in the prediction change
Prediction change to attributes on direction, including:
With reference to the overall prediction change direction, filtered out from the historical data and meet the overall prediction change side
To the first historical data;
Using first historical data, determine first parameter in the overall prediction change to attributes.
Alternatively, it is described to be based on the historical data, the prediction change direction in second time period is determined, including:
Based on the historical data, the first time period is split into multiple sub- periods;
The historical data corresponding to determining each described sub- period, determine the region of each sub- period
Predict change direction;
The regional prediction change direction with reference to described in and the historical data, determine first parameter in the prediction
Prediction change to attributes in change direction, including:
According to being corresponded in the second time period in the sub- period, pair of first parameter in the second time period is calculated
Answer in the sub- period, there is the predicted value of the regional prediction change to attributes.
Alternatively, the historical data corresponding to described each described sub- period of determination, determines each described son
The regional prediction change direction of period, including:
Wherein, the regional prediction change direction, including:Based on multiple discrete history in the sub- period each described
Data, count son prediction change direction corresponding to the two neighboring discrete historical data;
Son prediction change direction all in a sub- period is counted, selects the son prediction change of statistical value highest
Direction is as the regional prediction change direction.
Second aspect of the embodiment of the present invention provides a kind of electronic equipment, including:
Memory, for data;
Processor, it is connected with the memory, for the computer program by performing the memory storage, Neng Goushi
The data analysing method that existing said one or multiple skill schemes provide.
The third aspect of the embodiment of the present invention provides a kind of computer-readable storage medium, and the computer-readable storage medium is stored with meter
Calculation machine program, after the computer program is performed, the data analysis side that said one or multiple skill schemes provide can be realized
Method.
Data analysing method, electronic equipment and computer-readable storage medium provided in an embodiment of the present invention, in the first parameter
During predictor calculation, prediction change direction is can determine whether out first, then in conjunction with prediction change direction and historical data, is met
The predicted value of the prediction change direction.The predicted value so is calculated relative to being directly based upon historical data, due to head
First be predicted the determination of change direction, introduce the follow-up predictor calculation of more restrictions on the parameters, have accuracy high and
The characteristics of accuracy is high.
Brief description of the drawings
Fig. 1 is the schematic flow sheet of the first data analysing method provided in an embodiment of the present invention;
Fig. 2 is the schematic flow sheet of second of data analysing method provided in an embodiment of the present invention;
Fig. 3 is the structural representation of a kind of electronic equipment provided in an embodiment of the present invention;
Fig. 4 is the schematic flow sheet of the third data analysing method provided in an embodiment of the present invention.
Embodiment
Technical scheme is further elaborated below in conjunction with Figure of description and specific embodiment.
As shown in figure 1, the present embodiment provides a kind of data analysing method, including:
Step S110:Obtain the historical data of the first parameter in the first time period of history;
Step S120:Based on the historical data, the prediction change direction in second time period is determined, wherein, described
Two periods were later than the first time period;
Step S130:With reference to the prediction change direction and the historical data, the first parameter is calculated described second
In period, there is the predicted value of the prediction change direction.
The data analysing method that the present embodiment provides, can be applied to the information processing side of the server of various data analyses
Method.
First historical data of the first parameter correlation in the first time period that step S110 obtains history.Here
One period can be the period before current time.
Historical data is based in step S120, determines the prediction change direction in first time period.Here prediction becomes
Change direction, it may include:Rise change and decline change etc..
In the present embodiment, it is first determined go out change direction, then combining change direction and historical data, calculating the
One parameter has the predicted value of the prediction change direction in second time period.The pre- of change direction is predicted in having here
Measured value, it is that second time period relative to the change direction of first time period is the prediction change direction.
Using this data analysing method in the present embodiment, predicted value is calculated relative to historical data is directly based upon, extremely
Less by handling twice, obtain after predicting change direction, with reference to prediction change direction and historical data, calculate the pre- of the first parameter
Measured value, realize segmentation and calculate, the accuracy of calculating can be lifted.Change direction is predicted by combining, can be excluded indivedual big
The abnormal data of change also lifts the essence for carrying out data analysis by means of which from this aspect to the adverse effect of accuracy
Exactness.
The historical data in the present embodiment, can be various types of data, for example, selling commodity and/or service
Business datum, the day work amount data of application, the download installation data of application, the forwarding of article and/or comment data etc., in a word,
The historical data can be a variety of data, be not limited to any one.
Alternatively, as shown in Fig. 2 the step S130 specifically may include:
Step S131:With reference to the prediction change direction and the historical data, determine first parameter described pre-
The prediction change to attributes surveyed in change direction, wherein, the prediction change to attributes includes:Predict rate of change, prediction constant interval
And at least one of prediction rate of change grade;
Step S132:According to the prediction change to attributes and with reference to the historical data, the first parameter is calculated described
In second time period, there is the predicted value of the prediction change to attributes.
In the present embodiment, the step S130 is divided into two steps again, first, can be based on the prediction change direction and described
Historical data, determine prediction change to attributes of first parameter in default change direction.Here prediction change to attributes, specifically
Including with meet it is described prediction change direction prediction rate of change, the first parameter predicted value may prediction constant interval,
And/or prediction variation grades etc..
Determining the prediction change to attributes and then combining prediction change to attributes, and combining historical data, calculating
Predicted value with the prediction change to attributes.
For example, in certain embodiments, Trend Forecast corresponding to first parameter is growth trend, the prediction
Change to attributes, in a word, change to attributes is predicted described in the present embodiment, directly or indirectly give the predicted value of the first parameter described
Predict the corresponding section in change direction.
It is assumed that the prediction change direction is ascent direction, the prediction change to attributes is growth rate scope;It is then described pre-
Measured value is that before current time, growth rate is located at a prediction in the range of the growth rate in the ascent direction
Value.
In the present embodiment by the introducing of the prediction change to attributes, value is predicted again and finally calculates more ginsengs
Amount, so as to constrain the calculating of the predicted value from more perspective from so as to lift the final accuracy of the predicted value.
Alternatively, the step S131 may include:
With the first historical data in the historical data and the actual change direction of (n+1)th sub- period and actual become
Change attribute to be trained for training sample, so as to obtain attribute forecast model, wherein, first historical data includes:It is described
The 1st sub- period is to the historical data in n-th of sub- period in first time period;
Using the second historical data in the historical data by input data and to predict change direction, the category is input to
Property forecast model, the prediction change to attributes of the second time period is obtained, wherein, second historical data includes:2nd
The individual sub- period, n was the integer not less than 2 to the historical data in (n+1)th sub- period.
First historical data can be the actual value of first parameter, and the actual change direction can be described in utilization
First history parameters are by being calculated, for example, the actual change direction of (n+1)th sub- period, is to utilize (n+1)th son
The comparison of the initial data of the first parameter and the initial data of n-th of sub- period determines in period.(n+1)th son
The actual change attribute of period, it is same but sub using the initial data and n-th of the first parameter in (n+1)th sub- period
What the initial data of period was calculated.
The actual change direction and actual change category of the first historical data and (n+1)th sub-period are utilized in the present embodiment
Property, it is training sample, trains the attribute forecast model.For example, with first historical data and (n+1)th sub-period
Actual change direction be training sample input data;Using the actual change attribute as output result, neutral net is trained
Or learning machine or statistical regression model etc., obtain that the attribute forecast model for obtaining the prediction change to attributes can be carried out.
It is determined that it is described prediction change to attributes when, using described in the second historical data predict change direction as input data, it is defeated
Go out into the attribute forecast model trained, the attribute forecast model will export the prediction change to attributes naturally.
In certain embodiments, the step S132 may include:
With first historical data, actual change direction, actual change attribute and the institute of (n+1)th sub- period
The actual value for stating the first parameter carries out model training, so as to obtain value prediction model;
With second historical data, the prediction change direction, the prediction change to attributes are as input parameter, input
To the value prediction model, the predicted value of the second time period is obtained.
First with the first historical data, the actual change direction of (n+1)th sub- period and actual change attribute and first
The actual value of parameter carries out model training, obtains value prediction model.
When carrying out the training of described value training pattern, first historical data, the reality of (n+1)th sub- period
Border change direction, actual change attribute are the input data in training sample, and the actual value of first parameter is training pattern
Need output result, handled by repeated multiple times iteration etc., when the value prediction model, the first historical data of input, (n+1)th
The actual change direction of individual sub- period and actual change attribute, when can export the actual value of first parameter, so that it may think
It is to complete training.
In certain embodiments, can also be verified using sample is verified, if by checking, the attribute forecast model and
Value prediction model, it just can formally be used for the prediction for carrying out the predicted value of the first parameter of second time period.
Further, the step S110 may include:
Using the first historical data in the historical data and the prediction change direction of (n+1)th period as training sample
It is trained, so as to obtain direction prediction model, wherein, first historical data includes:The 1st in the first time period
The sub- period is to the historical data in n-th of sub- period
Using the second historical data in the historical data as input data, the directional prediction modes are input to, are obtained
Second time end the prediction change direction, wherein, second historical data includes:2nd sub- period is to n-th
The historical data in+1 sub- period, n are the integer not less than 2.
The prediction change direction in the present embodiment, determine again by corresponding model, become being predicted
Need to carry out model training before changing the prediction in direction, the model for now training to obtain is direction prediction model.
In certain embodiments, if first historical data is the actual value of N number of first parameter, the direction is trained
Forecast model, then the input parameter using the actual value as training sample, actual pre- to be obtained based on the calculated with actual values
Direction is surveyed, is that the output result of training pattern carries out model correction and repetitive exercise.
And when carrying out the attribute forecast model, then with N number of actual value, plus 1 actual change direction conduct
The input parameter of training sample, using the actual change attribute of (n+1)th sub-period as the output knot in training sample
Fruit, carry out model correction and repetitive exercise.Obviously, the parameter that the training of attribute forecast model uses is more more, constrained parameters
It is more more, it is clear that constraint is more, then all the more meets with actual conditions, then can be exported more using the model so trained
The predicted value of very near actual conditions.
Further, when carrying out the training of value predictive mode, then with N number of actual value, 1 actual change direction and actual change
Change attribute and carry out model training, and the comparison using actual value as the output result of model for the input parameter of training sample, use
Correction and repetitive exercise in model.Obviously, the input parameter in the training sample of value prediction model, is added, it is clear that about again
Beam parameter is again more, so as to again ensure that the accuracy of predicted value.
Above provide and utilize historical data training pattern, the method that prediction change direction is then drawn using model, with
It is lower that other several optional modes are provided:
Alternatively, the step S120 may also include:
The initial value of first parameter of initial time based on the first time period, and the first time period
The stop value of second parameter of end time;
Based on the initial value and the stop value, determine that the overall of the second time period predicts change direction.
The amount of money in the present embodiment utilizes the initial value of initial time, and the stop value of end time, first after it is whole
Body predicts change direction, and can directly ignore abnormity point causes what is occurred in very first time end to cause actual change direction ripple occur
It is dynamic.For example, an extremely low abnormity point may result in from growth trend and be changing into downward trend, change back growth trend again
Fluctuation.
The step S130 may include:
With reference to the overall prediction change direction, filtered out from the historical data and meet the overall prediction change side
To the first historical data;
Using first historical data, determine first parameter in the overall prediction change to attributes.
Overall prediction direction is recycled, the abnormity point for causing direction to be fluctuated is filtered out, for example, above-mentioned extremely low exception
Point, the filtering of exceptional value is so obviously achieved that, so as to the inaccurate problem of predicted value caused by avoiding exceptional value.
Alternatively, the step S120 may also include:
Based on the historical data, the first time period is split into multiple sub- periods;
The historical data corresponding to determining each described sub- period, determine the region of each sub- period
Predict change direction;
The step S130 may include:
According to being corresponded in the second time period in the sub- period, pair of first parameter in the second time period is calculated
Answer in the sub- period, there is the predicted value of the regional prediction change to attributes.
For example, in the present embodiment, the first time period can also be split into multiple sub- periods, be then based on history
Data, obtain the regional prediction change direction of each sub- period.
Then the predicted value is obtained to the sub- time end of one according to second time period.
Alternatively, the step S110 may also include:
Wherein, the regional prediction change direction, including:Based on multiple discrete history in the sub- period each described
Data, count son prediction change direction corresponding to the two neighboring discrete historical data;
Son prediction change direction all in a sub- period is counted, selects the son prediction change of statistical value highest
Direction is as the regional prediction change direction.
No longer only it is two end values of initial time and end time based on first time period in the present embodiment to carry out
Calculate, but first time period is split into multiple sub- time ends, then obtain the sub- change direction of each sub- period, so
Afterwards based on statistics, the son prediction change direction more than comparison is obtained, as final prediction change direction.If overall variation rises
, even if then obviously splitting into multiple sub- periods, then the son prediction change direction of sub- period, is in the great majority.Based on this
Kind mode, can also exclude the problem of exceptional value causes.
In certain embodiments, the step S131 may include:
Compare the first historical data and the second historical data of the second sub- period of the first time period;
Based on the second historical data and the first historical data, when calculating for the described second sub- period relative to described first
Between in section first parameter change ratio on the history direction;
Determine the variation grades corresponding to the change ratio.
As shown in figure 3, the present embodiment provides a kind of electronic equipment, including:
Memory 110, for data;
Processor 120, it is connected with the memory, for the computer program stored by performing the memory 110,
The data analysing method of the offer of one or more technical schemes can be provided.
The memory 110 may include various types of storage mediums, foregoing available for storing various data, such as code
Historical data, the related data of various models, and storage medium etc..The storage medium that the memory 110 includes, at least
Part is non-moment storage medium, and the non-moment storage medium can be used for storing computer program.
The memory processor can be central processing unit, microprocessor, digital signal processor, programmable array or answer
With processor etc..The processor can also be application specific integrated circuit etc..
The processor 120 can be connected, for reading by EBIs such as IC bus with the memory 110
The computer program in the memory 110 is taken, and foregoing one or more technical sides are realized by performing the computer program
The data analysing method that case provides.
The present embodiment also provides a kind of computer-readable storage medium, and the computer-readable storage medium is stored with computer program,
After the computer program is performed, the data analysing method of the offer of one or more technical schemes can be realized.
Here computer-readable storage medium can be:Movable storage device, read-only storage (ROM, Read-Only
Memory), random access memory (RAM, Random Access Memory), magnetic disc or CD etc. are various can store journey
The medium of sequence code, it is chosen as non-moment storage medium.
The embodiment of the present invention also provides a kind of server, including:
Acquiring unit, the historical data of the first parameter in the first time period for obtaining history;The acquiring unit, can
Communication interface is may correspond to, is obtained from peripheral hardware;Or corresponding to processor, read from local storage medium, be used to obtain in a word
Take historical data
Determining unit, for based on the historical data, determining the prediction change direction in second time period, wherein, institute
State second time period and be later than the first time period;Here processor may correspond to processor or process circuit etc.;
Computing unit, for reference to the prediction change direction and the historical data, calculating the first parameter described
In second time period, there is the predicted value of the prediction change direction.Computing unit may correspond to calculator or with calculating work(
Processor of energy etc..
Alternatively, the computing unit, specifically for reference to the prediction change direction and the historical data, determining institute
Prediction change to attributes of first parameter in the prediction change direction is stated, wherein, the prediction change to attributes includes:Prediction becomes
At least one of rate, prediction constant interval and prediction rate of change grade;According to the prediction change to attributes and combine institute
Historical data is stated, calculates the first parameter in the second time period, the predicted value with the prediction change to attributes.
Alternatively, the computing unit, also particularly useful for the first historical data in the historical data and (n+1)th
The actual change direction of sub- period and actual change attribute are trained for training sample, so as to obtain attribute forecast model,
Wherein, first historical data includes:The 1st sub- period is to the institute in n-th of sub- period in the first time period
State historical data;Using the second historical data in the historical data by input data and to predict change direction, institute is input to
Attribute forecast model is stated, obtains the prediction change to attributes of the second time period, wherein, the second historical data bag
Include:2nd sub- period, n was the integer not less than 2 to the historical data in (n+1)th sub- period.
Alternatively, computing unit, for the actual change of first historical data, (n+1)th sub- period
The actual value in direction, actual change attribute and first parameter carries out model training, so as to obtain value prediction model;With described
Second historical data, the prediction change direction, the prediction change to attributes are input to described value prediction mould as input parameter
Type, obtain the predicted value of the second time period.
In certain embodiments, the computing unit, it is additionally operable to the first historical data and n-th in the historical data
The prediction change direction of+1 period is trained for training sample, so as to obtain direction prediction model, wherein, described first
Historical data includes:The 1st sub- period is to the historical data in n-th of sub- period in the first time period;With
The second historical data in the historical data is input data, the directional prediction modes is input to, when obtaining described second
Between the prediction change direction held;Wherein, second historical data includes:The 2nd sub- period to (n+1)th sub- time
The historical data in section, n are the integer not less than 2.
Further, the determining unit, specifically for described first of the initial time based on the first time period
The initial value of parameter, and the stop value of the second parameter of the end time of the first time period;Based on the initial value and institute
Stop value is stated, determines the overall prediction change direction of the second time period;The computing unit, specifically for reference to described whole
Body predicts change direction, and the first historical data for meeting the overall prediction change direction is filtered out from the historical data;
Using first historical data, determine first parameter in the overall prediction change to attributes.
In addition, the determining unit, specifically for based on the historical data, the first time period being split into multiple
The sub- period;The historical data corresponding to determining each described sub- period, determine the area of each sub- period
Predict change direction in domain;The computing unit, it is additionally operable to correspond in the sub- period according in the second time period, calculates
One parameter has the predicted value of the regional prediction change to attributes within the correspondence sub- period of the second time period.
In certain embodiments, the determining unit, it is additionally operable to wherein, the regional prediction change direction, including:It is based on
Multiple discrete historical datas in each described sub- period, count sub corresponding to the two neighboring discrete historical data
Predict change direction;Son prediction change direction all in a sub- period is counted, selects statistical value highest pre-
Change direction is surveyed as the regional prediction change direction.
Below in conjunction with any one above-mentioned embodiment, there is provided several specific examples:
Example 1:
This example provides a kind of data analysing method, including:
The first step:The sales volume for predicting next day according to historical data is rise or drop, by problem reduction into one
Two classification problems, it is predicted using two disaggregated models.
Second step:According to the result of the first step, then predict that next day sales volume goes up or dropped compared with the previous day sales volume
Ratio (ratio is set to several grades), problem is reduced to more classification problems from the specific numerical value for going up or declining of prediction, used
More classification problems are predicted.
3rd step:According to the result of the first two steps, in the candidate section of diminution, finally next day sales volume is gone up
Or the numerical value of drop is predicted, and is predicted using regression model.
Compared with the conventional method, this method has the advantage that:
The problem of direct prediction sales volume, is disassembled and from big to small, is solved one by one for three steps, granularity, to lift prediction
Accuracy rate
By the way that prediction to change direction, rate of change will be converted into the prediction of concrete numerical value, make model more for robust
Property, there is certain fault-tolerance to exceptional value, it is not easy to over-fitting
Example 2:
This hair example provides a kind of data analysing method, first, is carried out by the rise or drop of next day sales volume pre-
Survey, determine change direction;Secondly, variation grades are divided by the prediction to rate of change, then in obtained change
The specific changing value of sales volume is predicted in grade, so as to finally give the predicted value of next day sales volume.
This example considers the change direction of next day sales volume, rate of change and specific changing value, so as to enter to problem
Row decomposes and obtains the predicted value of next day sales volume, and its meaning is, the present invention compared with prior art, can not only improve
The quality of prediction, and have certain fault-tolerance to exceptional value, over-fitting is not easy, the robustness of model is greatly improved.
Example 3:
As shown in figure 4, this example provides a kind of data analysing method, including:
Step S1:Obtain the historical data of sales volume;
Step S2:It is up-trend or downward tendency to predict next day sales volume;
Step S3:Predict the rate of change grade in corresponding trend;
Step S4:Predict that next day meets the sales volume of the rate of change grade.
In several embodiments provided herein, it should be understood that disclosed apparatus and method, it can be passed through
Its mode is realized.Apparatus embodiments described above are only schematical, for example, the division of the unit, is only
A kind of division of logic function, there can be other dividing mode when actually realizing, such as:Multiple units or component can combine, or
Another system is desirably integrated into, or some features can be ignored, or do not perform.In addition, shown or discussed each composition portion
Point mutual coupling or direct-coupling or communication connection can be the INDIRECT COUPLINGs by some interfaces, equipment or unit
Or communication connection, can be electrical, mechanical or other forms.
The above-mentioned unit illustrated as separating component can be or may not be physically separate, show as unit
The part shown can be or may not be physical location, you can positioned at a place, can also be distributed to multiple network lists
In member;Partly or entirely unit therein can be selected to realize the purpose of this embodiment scheme according to the actual needs.
In addition, each functional unit in various embodiments of the present invention can be fully integrated into a processing module, also may be used
To be each unit individually as a unit, can also two or more units it is integrated in a unit;It is above-mentioned
Integrated unit can both be realized in the form of hardware, can also be realized in the form of hardware adds SFU software functional unit.
One of ordinary skill in the art will appreciate that:Realizing all or part of step of above method embodiment can pass through
Programmed instruction related hardware is completed, and foregoing program can be stored in a computer read/write memory medium, the program
Upon execution, the step of execution includes above method embodiment.
The foregoing is only a specific embodiment of the invention, but protection scope of the present invention is not limited thereto, any
Those familiar with the art the invention discloses technical scope in, change or replacement can be readily occurred in, should all be contained
Cover within protection scope of the present invention.Therefore, protection scope of the present invention should be based on the protection scope of the described claims.
Claims (10)
- A kind of 1. data analysing method, it is characterised in that including:Obtain the historical data of the first parameter in the first time period of history;Based on the historical data, the prediction change direction in second time period is determined, wherein, the second time period is later than institute State first time period;With reference to the prediction change direction and the historical data, the first parameter is calculated in the second time period, is had The predicted value of the prediction change direction.
- 2. according to the method for claim 1, it is characterised in thatIt is described that change direction and the historical data are predicted with reference to described in, the first parameter is calculated in the second time period, Predicted value with the prediction change direction, including:With reference to the prediction change direction and the historical data, determine first parameter in the prediction change direction Change to attributes is predicted, wherein, the prediction change to attributes includes:Predict rate of change, prediction constant interval and prediction rate of change etc. At least one of level;According to the prediction change to attributes and with reference to the historical data, the first parameter is calculated in the second time period, Predicted value with the prediction change to attributes.
- 3. according to the method for claim 2, it is characterised in thatIt is described that change direction and the historical data are predicted with reference to described in, determine first parameter in the prediction change direction On prediction change to attributes, including:With the actual change direction and actual change category of the first historical data in the historical data and (n+1)th sub- period Property is trained for training sample, so as to obtain attribute forecast model, wherein, first historical data includes:Described first The 1st sub- period is to the historical data in n-th of sub- period in period;Using the second historical data in the historical data by input data and to predict change direction, it is pre- to be input to the attribute Model is surveyed, obtains the prediction change to attributes of the second time period, wherein, second historical data includes:2nd son Period, n was the integer not less than 2 to the historical data in (n+1)th sub- period.
- 4. according to the method for claim 3, it is characterised in thatIt is described to predict change to attributes and with reference to the historical data according to described, the first parameter is calculated in the second time period It is interior, there is the predicted value of the prediction change to attributes, including:With first historical data, the actual change direction of (n+1)th sub- period, actual change attribute and described The actual value of one parameter carries out model training, so as to obtain value prediction model;With second historical data, the prediction change direction, the prediction change to attributes are input to institute as input parameter Value prediction model is stated, obtains the predicted value of the second time period.
- 5. according to the method described in any one of Claims 1-4, it is characterised in thatIt is described to be based on the historical data, the prediction change direction in second time period is determined, including:Carried out using the first historical data in the historical data and the prediction change direction of (n+1)th period as training sample Training, so as to obtain direction prediction model, wherein, first historical data includes:1st period of the day from 11 p.m. to 1 a.m in the first time period Between section to the historical data in n-th of sub- period;Using the second historical data in the historical data as input data, the directional prediction modes are input to, are obtained described Second the time end the prediction change direction;Wherein, second historical data includes:2nd sub- period is to (n+1)th The historical data in the sub- period, n are the integer not less than 2.
- 6. according to the method for claim 1, it is characterised in thatIt is described to be based on the historical data, the prediction change direction in second time period is determined, including:The initial value of first parameter of initial time based on the first time period, and the termination of the first time period The stop value of second parameter at moment;Based on the initial value and the stop value, determine that the overall of the second time period predicts change direction;It is described that change direction and the historical data are predicted with reference to described in, determine first parameter in the prediction change direction On prediction change to attributes, including:With reference to the overall prediction change direction, filtered out from the historical data and meet the overall prediction change direction First historical data;Using first historical data, determine first parameter in the overall prediction change to attributes.
- 7. according to the method described in any one of Claims 1-4, it is characterised in thatIt is described to be based on the historical data, the prediction change direction in second time period is determined, including:Based on the historical data, the first time period is split into multiple sub- periods;The historical data corresponding to determining each described sub- period, determine the regional prediction of each sub- period Change direction;The regional prediction change direction with reference to described in and the historical data, determine first parameter in the prediction change Prediction change to attributes on direction, including:According to being corresponded in the second time period in the sub- period, corresponding son of first parameter in the second time period is calculated In period, there is the predicted value of the regional prediction change to attributes.
- 8. according to the method for claim 7, it is characterised in thatThe historical data corresponding to described each described sub- period of determination, determine the region of each sub- period Change direction is predicted, including:Wherein, the regional prediction change direction, including:Based on multiple discrete history numbers in the sub- period each described According to counting corresponding to the two neighboring discrete historical data son prediction change direction;Son prediction change direction all in a sub- period is counted, selects statistical value highest son prediction change direction As the regional prediction change direction.
- 9. a kind of electronic equipment, it is characterised in that including:Memory, for data;Processor, it is connected with the memory, for the computer program by performing the memory storage, power can be realized Profit requires the method described in 1 to 8 any one.
- 10. a kind of computer-readable storage medium, the computer-readable storage medium is stored with computer program, the computer program quilt After execution, the method described in any one of claim 1 to 8 can be realized.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710488327.1A CN107357764B (en) | 2017-06-23 | 2017-06-23 | Data analysis method, electronic device, and computer storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710488327.1A CN107357764B (en) | 2017-06-23 | 2017-06-23 | Data analysis method, electronic device, and computer storage medium |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107357764A true CN107357764A (en) | 2017-11-17 |
CN107357764B CN107357764B (en) | 2020-10-27 |
Family
ID=60272456
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710488327.1A Active CN107357764B (en) | 2017-06-23 | 2017-06-23 | Data analysis method, electronic device, and computer storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107357764B (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108921323A (en) * | 2018-06-04 | 2018-11-30 | 北京百度网讯科技有限公司 | Method and apparatus for generating information |
CN110019368A (en) * | 2017-12-29 | 2019-07-16 | 西门子(中国)有限公司 | Industrial data analysis method and device, computer storage medium |
CN110059069A (en) * | 2018-01-19 | 2019-07-26 | 西门子股份公司 | System and method for detecting and predicting the behavior of goal systems |
WO2020037511A1 (en) * | 2018-08-21 | 2020-02-27 | 华为技术有限公司 | Data storage and acquisition method and device |
CN112561121A (en) * | 2020-11-12 | 2021-03-26 | 北京大学 | Rework trend prediction method and system based on mobile phone signaling data |
CN112990552A (en) * | 2021-02-20 | 2021-06-18 | 节点互联(北京)科技有限公司 | Equipment operation parameter short-time prediction method and system based on change rate |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105205297A (en) * | 2014-06-12 | 2015-12-30 | 腾讯科技(深圳)有限公司 | Time series prediction method and system |
CN105303835A (en) * | 2015-11-13 | 2016-02-03 | 西安邮电大学 | Short-time prediction method of road traffic flow state |
CN105956122A (en) * | 2016-05-03 | 2016-09-21 | 无锡雅座在线科技发展有限公司 | Object attribute determining method and device |
-
2017
- 2017-06-23 CN CN201710488327.1A patent/CN107357764B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105205297A (en) * | 2014-06-12 | 2015-12-30 | 腾讯科技(深圳)有限公司 | Time series prediction method and system |
CN105303835A (en) * | 2015-11-13 | 2016-02-03 | 西安邮电大学 | Short-time prediction method of road traffic flow state |
CN105956122A (en) * | 2016-05-03 | 2016-09-21 | 无锡雅座在线科技发展有限公司 | Object attribute determining method and device |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110019368A (en) * | 2017-12-29 | 2019-07-16 | 西门子(中国)有限公司 | Industrial data analysis method and device, computer storage medium |
CN110059069A (en) * | 2018-01-19 | 2019-07-26 | 西门子股份公司 | System and method for detecting and predicting the behavior of goal systems |
CN108921323A (en) * | 2018-06-04 | 2018-11-30 | 北京百度网讯科技有限公司 | Method and apparatus for generating information |
WO2020037511A1 (en) * | 2018-08-21 | 2020-02-27 | 华为技术有限公司 | Data storage and acquisition method and device |
JP2021534505A (en) * | 2018-08-21 | 2021-12-09 | 華為技術有限公司Huawei Technologies Co., Ltd. | Data storage method, data acquisition method, and equipment |
JP7108784B2 (en) | 2018-08-21 | 2022-07-28 | 華為技術有限公司 | DATA STORAGE METHOD, DATA ACQUISITION METHOD, AND DEVICE |
US11960467B2 (en) | 2018-08-21 | 2024-04-16 | Huawei Technologies Co., Ltd. | Data storage method, data obtaining method, and apparatus |
CN112561121A (en) * | 2020-11-12 | 2021-03-26 | 北京大学 | Rework trend prediction method and system based on mobile phone signaling data |
CN112561121B (en) * | 2020-11-12 | 2021-07-20 | 北京大学 | Rework trend prediction method and system based on mobile phone signaling data |
CN112990552A (en) * | 2021-02-20 | 2021-06-18 | 节点互联(北京)科技有限公司 | Equipment operation parameter short-time prediction method and system based on change rate |
Also Published As
Publication number | Publication date |
---|---|
CN107357764B (en) | 2020-10-27 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107357764A (en) | Data analysing method, electronic equipment and computer-readable storage medium | |
CN109242135B (en) | Model operation method, device and business server | |
CN104915735B (en) | The early warning analysis method of enterprise operation analyzing and alarming system | |
CN108764974A (en) | A kind of procurement of commodities amount prediction technique and device based on deep learning | |
CN107609289B (en) | Building material cost control method and system for structural reinforcement engineering based on BIM model | |
KR101171539B1 (en) | Model optimization system using variable scoring | |
CN108960719A (en) | Selection method and apparatus and computer readable storage medium | |
US20110010226A1 (en) | Marketing model determination system | |
CN108734330A (en) | Data processing method and device | |
CN107480028A (en) | The acquisition methods and device of residual time length workable for disk | |
CN108256898A (en) | A kind of product Method for Sales Forecast method, system and storage medium | |
CN107392645A (en) | Usage mining method, apparatus and its equipment | |
CN107491907A (en) | Cost for Electric Power Engineering control and analysis and system | |
CN107908536A (en) | To the performance estimating method and system of GPU applications in CPU GPU isomerous environments | |
CN114048436A (en) | Construction method and construction device for forecasting enterprise financial data model | |
CN108491991A (en) | Constraints analysis system based on the industrial big data product duration and method | |
CN110288460A (en) | Collection prediction technique, device, equipment and storage medium based on propagated forward | |
CN108256802B (en) | Crowd search algorithm-based multi-supplier order distribution cloud processing method | |
CN106528774A (en) | Method and apparatus for predicting distribution network project management trend | |
CN112150179B (en) | Information pushing method and device | |
CN115034812B (en) | Steel industry sales volume prediction method and device based on big data | |
CN111984637B (en) | Missing value processing method and device in data modeling, equipment and storage medium | |
CN109784524A (en) | Part amount prediction technique and device, equipment and storage medium | |
CN107291722B (en) | Descriptor classification method and device | |
JP2005157836A (en) | Device and method for predicting merchandise sales and method for learning model |
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 |