CN108665113A - Index prediction technique and device - Google Patents
Index prediction technique and device Download PDFInfo
- Publication number
- CN108665113A CN108665113A CN201810483404.9A CN201810483404A CN108665113A CN 108665113 A CN108665113 A CN 108665113A CN 201810483404 A CN201810483404 A CN 201810483404A CN 108665113 A CN108665113 A CN 108665113A
- Authority
- CN
- China
- Prior art keywords
- window
- index
- time series
- prediction
- specified quantity
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- 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"
Abstract
Disclose a kind of index prediction technique and device.This method includes：The time series compared with isometric multiple of each of multiple second windows respectively is determined in the historical time sequence of first window；In the multiple relatively time series, the comparison time series with the most similar specified quantity of each base time series is selected respectively；According to the prediction result for determining third Window Index respectively with the comparison time series of the most similar specified quantity of each base time series；The accuracy rate of each prediction result is determined according to the historical time sequence, and selects the highest prediction result of accuracy rate for the final prediction result of the third Window Index.Influence of the benchmark length of window to prediction result is reduced with this, to reduce the requirement to benchmark window and the trend consistency of historical time sequence, improves the accuracy of prediction result.
Description
Technical field
This specification is related to field of computer technology more particularly to a kind of index prediction technique and device.
Background technology
With the development of science and technology and the aggravation of industry competition, the control of cost, risk etc. at increase enterprise key
The necessary means of competitiveness can reduce cost and risk for the accurately prediction of operational indicator future trend.
The classical time series algorithm of traditional operational indicator prediction use, such as method of moving average, sliding average,
ARIMA or Holt-Winters etc., it is time series numerical value to enter ginseng, and provides future services index according to different algorithms
Development trend.But these time series algorithms are higher to the coherence request of time series trend, i.e., if nearest industry
Business development trend has exception, then the prediction result obtained according to this is also probably abnormal.
Invention content
In view of the above technical problems, a kind of index prediction technique of this specification offer and device.
Specifically, this specification is achieved by the following technical solution：
In a first aspect, this specification embodiment provides a kind of index prediction technique.This method includes：
Multiple ratios isometric with each of multiple second windows respectively are determined in the historical time sequence of first window
Compared with time series, each second window corresponds to a base time series, and each second window corresponds to different length of window, described
The length of window of first window is more than second window, and the first window is identical as the terminal of the second window；
The multiple relatively in time series, the ratio of selection and the most similar specified quantity of each base time series respectively
Compared with time series；
Third window is determined respectively according to the comparison time series with the most similar specified quantity of each base time series
The prediction result of mouth index, the starting point of the third window are adjacent with the terminal of the second window；
The accuracy rate of each prediction result is determined according to the historical time sequence, and selects the highest prediction knot of accuracy rate
Fruit is the final prediction result of the third Window Index.
Second aspect, this specification embodiment provide a kind of index prediction meanss, which is characterized in that the device includes：
Determination unit, for being determined in the historical time sequence of first window respectively each of with multiple second windows
Isometric multiple relatively time serieses, each second window correspond to a base time series, and each second window corresponds to different
The length of window of length of window, the first window is more than second window, the first window and second window
Terminal is identical；
First selecting unit, in the multiple relatively time series, selecting respectively with each base time series most
The comparison time series of similar specified quantity；
Predicting unit, for according to the comparison time series point with the most similar specified quantity of each base time series
Not Que Ding third Window Index prediction result, the starting point of the third window is adjacent with the terminal of the second window；
Second selecting unit for determining the accuracy rate of each prediction result according to the historical time sequence, and selects
The highest prediction result of accuracy rate is the final prediction result of the third Window Index.
The third aspect, this specification embodiment provide a kind of computer equipment, including memory, processor and are stored in
On memory and the computer program that can run on a processor, which is characterized in that the processor executes real when described program
The method and step of existing aforementioned first aspect.
Fourth aspect provides a kind of computer readable storage medium, and meter is stored on the computer readable storage medium
Calculation machine program, the computer program realize the method described in above-mentioned first aspect when being executed by processor.
5th aspect provides a kind of computer program product including instruction, when described instruction is run on computers
When so that computer executes the method described in above-mentioned first aspect.
By this specification embodiment, benchmark is used as by different length of window, future services index is predicted,
And historical data is combined to calculate the corresponding predictablity rate of each length of window, select the highest window of accuracy rate corresponding with this
Prediction result reduces influence of the benchmark length of window to prediction result, to reduce to benchmark window and historical time sequence
The requirement of trend consistency improves the accuracy of prediction result.
It should be understood that above general description and following detailed description is only exemplary and explanatory, not
The application this specification embodiment can be limited.
In addition, any embodiment in the application this specification embodiment does not need to reach above-mentioned whole effects.
Description of the drawings
In order to illustrate more clearly of this specification embodiment or technical solution in the prior art, below will to embodiment or
Attached drawing needed to be used in the description of the prior art is briefly described, it should be apparent that, the accompanying drawings in the following description is only
Some embodiments described in this specification embodiment for those of ordinary skill in the art can also be attached according to these
Figure obtains other attached drawings.
Fig. 1 is that this specification implements a kind of application scenarios schematic diagram exemplified；
Fig. 2 is the flow diagram that this specification implements a kind of index prediction technique exemplified；
Fig. 3 is that this specification implements the example exemplified；
Fig. 4 is the flow diagram that this specification implements another index prediction technique exemplified；
Fig. 5 is the structural schematic diagram that this specification implements a kind of index prediction meanss exemplified；
Fig. 6 is the structural schematic diagram that this specification implements a kind of computer equipment exemplified.
Specific implementation mode
Example embodiments are described in detail here, and the example is illustrated in the accompanying drawings.Following description is related to
When attached drawing, unless otherwise indicated, the same numbers in different drawings indicate the same or similar elements.Following exemplary embodiment
Described in embodiment do not represent all embodiments consistent with this specification.On the contrary, they are only and such as institute
The example of the consistent device and method of some aspects be described in detail in attached claims, this specification.
It is the purpose only merely for description specific embodiment in the term that this specification uses, is not intended to be limiting this explanation
Book.The "an" of used singulative, " described " and "the" are also intended to packet in this specification and in the appended claims
Most forms are included, unless context clearly shows that other meanings.It is also understood that term "and/or" used herein is
Refer to and include one or more associated list items purposes any or all may combine.Depending on context, as made at this
Word " if " can be construed to " ... when " or " when ... " or " in response to determination ".
With the development of science and technology and the aggravation of industry competition, the control of cost, risk etc. at increase enterprise key
The necessary means of competitiveness can reduce cost and risk for the accurately prediction of operational indicator future trend.
By taking Alipay international payment business as an example.Alipay provides the consumption abroad clothes on line and under line to client at present
It is engaged in, the online shopping of website is washed in a pan in the sea such as international and similar koala nets of service You Tianmao on line, has Hong Kong, South Korea, day under line
Sheet, Southeast Asia and some other country overseas pay service face to face.Either on line or under line, client only needs payer
People's coin, but Alipay needs to settle accounts corresponding foreign currency to different businessmans, therefore there is demand of largely exchanging in Alipay.In order to the greatest extent
Amount reduces influence of the fluctuation of exchange rate to Alipay, purchase of the Alipay in need of work that morning just and on the day of counterparty's locking
Foreign exchange volume, in addition, since in weekend or festivals or holidays, open window is not traded for bank and other financial mechanism counterparty side, therefore
Such as weekend or festivals or holidays, need to be purchased in advance weekend in advance on the last one working day/festivals or holidays needed for foreign exchange, therefore it is accurately pre-
Survey business occurrence quantity, it is possible to reduce business risk and raising fund utilization ratio.
In view of the above problems, a kind of index prediction technique of this specification embodiment offer and device, first below to this theory
The operating system framework of bright book example scheme illustrates.Entity shown in Figure 1, that this specification example scheme is related to
Including：Computer equipment 100 and terminal 200, wherein terminal 200 is generated for being used in specific business scenario
Business datum, the business datum include operational indicator, for example, overseas being paid using mobile phone, are carried out using laptop
Abroad buy on behalf etc..Computer equipment 100 is used to that terminal 200 to be supported to provide corresponding business, and is carried out centainly to terminal 200
Management.Such as the business datum of the generation of each terminal 200 can be collected, it is managed and is formed through over cleaning etc. according to the business datum
Historical time sequence is predicted according to the index in the historical time sequence pair future.
The embodiment of the present invention is further introduced below in conjunction with the accompanying drawings.
Fig. 2 is the flow diagram for the index prediction technique that this specification embodiment provides.This method is suitable for computer
Equipment, as shown in Fig. 2, this method tool includes step 210-240：
S210, determined in the historical time sequence of first window respectively with isometric more of each of multiple second windows
A relatively time series.Wherein, each second window corresponds to a base time series, and each second window corresponds to different windows
The length of window of length, the first window is more than second window, the terminal of the first window and second window
It is identical.In addition, time series is the time series of index.
It is foundation using first window middle finger target historical time sequence, with second in the scheme that this specification is provided
In window on the basis of base time series, the index in third window is predicted.
In conjunction with shown in Fig. 3, first window is historical data value window, which can be time span or number
According to amount length, value can determine according to actual needs, for example, time series is the index daily counted, the first window
Starting point was before three months, and terminal is current time, and length of window is three months.Alternatively, when the starting point of the first window is current
The 60th index of history is carved, terminal is current time index, length of window 60.In addition, the first window can also be covered
All historical datas.For another example first window is 14 achievement datas shown in Fig. 3.
In one example, it can first determine that source data, the source data are history service data.The source data is carried out
Cleaning and processing, generate historical time sequence.When can select the history that first window includes in the historical time sequence
Between the foundation predicted as index of sequence.
Second window is the value window of base time series, and the benchmark which predicts as index becomes
Gesture.Wherein, second window is identical as the website of first window, and the second window is included in first window.For example, given the
After two length of window, the base time sequence of last continuous several indexs compositions of historical time sequence in first window is generally taken
Row.For example, first window shown in Fig. 3 is moment 301 to the moment 314, and the second window can be moment 311 to the moment 314.
Third window is the window for the index for needing to predict, the terminal phase of the starting point of the third window and second window
It is adjacent.For example, the terminal of the second window is current time, then the starting point of third window is subsequent time, alternatively, the second window
Terminal is last moment, then the starting point of third window is current time.The value of the third window determines according to actual needs,
For example, the third window can be two days or two, one day or one following.For example, first window is the moment as shown in Figure 3
301 to the moment 314, and the second window can be moment 311 to the moment 314, third window is moment 315 to the moment 316.
In this specification embodiment, the second window includes multiple, each corresponds to different length, specifically, Ke Yishe
The value range of the length of fixed second window, when being predicted into row index, it is thus necessary to determine that all possible in the value range to take
Second window of value.Wherein it is possible to determine the value range according to actual needs.For example, according to the need of index predictablity rate
It asks, the second length of window is typically no less than 3 data, and maximum is not more than 10 data.For example, in conjunction with shown in Fig. 3, if the second window
Mouthful value range be not less than 3, then the second window can be moment 312 to the moment 314, moment 311 to the moment 314, when
Carve 310 to the moment 314, moment 309 to the moment 314, moment 308 to the moment 314, moment 307 to the moment 314, the moment 306 to when
Carve 314, moment 305 to the moment 314, moment 304 to the moment 314, moment 303 to the moment 314, moment 302 to the moment 314 and when
Carve 301 to the moment 314.
Further, it when being predicted into row index, for each of multiple second windows, needs in historical time sequence
Middle multiple relatively time series, the comparison time series terminals for selecting length to be equal to the second length of window for length of window are adjacent
There is history index, the quantity of the adjacent history index is more than or equal to the length of third window, for example, in conjunction with shown in Fig. 3, second
When window is moment 311 to the moment 314, when needing to predict the index at lower two moment, compare time series include the moment 301 to
Any four continuous moment corresponding all time serieses in moment 312, to compare time series as moment 309 to the moment 312
For corresponding time series, the adjacent history index of the comparison time series terminal include moment 313 and moment 314 it is corresponding
Index.It is selected from the historical time sequence of first window generally in a manner of traversal.For example, first window length is 60, the
Two length of window are 3, and third length of window is 1, then, multiple relatively time serieses which is 3 are that length is 60 to go through
All combinations in history time series in all continuous 3 indexs in addition to corresponding continuous three indexs of the second window.
S220 is selected and the most similar specified number of each base time series respectively in the multiple relatively time series
The comparison time series of amount.Wherein, specified quantity is the positive integer more than or equal to 1.
For each of multiple second windows, can be selected most close in its corresponding multiple relatively time series
Specified quantity comparison time series, which can determine according to actual needs.For example, the specified quantity can be
10,20 or 30 etc..Wherein, the determination of base time series and the similarity for comparing time series may include various ways,
Such as can be determined by calculating the similarity between the vector that it is respectively formed, it can specifically be achieved by the steps of：
A base vector is determined respectively according to each base time series, and each base vector is by its corresponding Ji Shijianxuliebao
The index included is constituted；
Determine that a comparison is vectorial according to each relatively time series, it is each more vectorial by its corresponding relatively time sequence
The index that row include is constituted；
According to base vector and more vectorial similarity, determine comparison with the most similar specified quantity of each base vector to
Amount, wherein the comparison of each most similar specified quantity of base vector it is vectorial by with the most similar specified number of each base time series
The comparison time series of amount is constituted.
Wherein, the standard that is described as similarity of cosine angle between vector, it is vectorial between cosine angle to it is similar
Degree is negatively correlated.Calculate that vector is similar in addition to co sinus vector included angle value mentioned above, can also by the Euclidean distance of vector with
And absolute value of the difference of vector etc. movement description.
S230 determines the respectively according to the comparison time series with the most similar specified quantity of each base time series
The prediction result of three Window Index.Wherein, the starting point of the third window is adjacent with the terminal of the second window.
Wherein, third window is determined according to the comparison time series with the most similar specified quantity of each base time series respectively
The prediction result of mouth index at least can be in the following way：
In one example, prediction index can be determined respectively according to time series is compared, after then eliminating exceptional value,
Synthesis obtains prediction result, is based on this, this method can be achieved by the steps of：
Specified quantity is determined respectively according to the comparison time series with the most similar specified quantity of each base time series
The prediction index of third Window Index；
Abnormal index in the corresponding specified quantity prediction index of each second window is rejected；
Determine the prediction result of third Window Index respectively according to the prediction index after rejecting.
Wherein, the adjacent history index of the terminal of time series may be considered the comparison time series trend compared with each
Corresponding actual value can be adjusted the actual value to obtain predicted value, for example, in conjunction with shown in Fig. 3, comparing time series is
When 312 corresponding time series of moment 309 to moment, multiple continuous history indexs are 314 corresponding finger of moment 313 and moment
Mark.The moment 313 and 314 corresponding index of moment are that the corresponding relatively time series trend of moment 309 to moment 312 corresponds to
Actual value.Based on this, the prediction index of third Window Index can determine as follows：
Determine the adjacent multiple continuous history indexs identical with third length of window of terminal of each relatively time series；
Proportion of utilization method is adjusted each history index, respectively obtains the corresponding specified quantity of each second window
The prediction index of third Window Index.
Wherein, rule of three is specifically as follows predicted value=actually occur value * Avg (base vector/more vectorial), and Avg is indicated
Mean function.Base vector is the corresponding vector of base time series, and more vectorial is to compare the corresponding vector of time series.
In another example, the threshold value that prediction index can be predefined according to historical time sequence, exceeds all
The prediction index of the threshold value is adjusted within the scope of this, then determines prediction result according to the prediction index after adjustment.Specifically, should
Adjustment process can be combined with aforementioned rejecting process, more accurately be predicted with realizing.Based on this, prediction result can be by as follows
Step determines：
Specified quantity is determined respectively according to the comparison time series with the most similar specified quantity of each base time series
The prediction index of third Window Index；
Abnormal index in the corresponding specified quantity prediction index of each second window is rejected；
The prediction index for exceeding the threshold value in prediction index after rejecting is adjusted in the threshold value；
Prediction index after being adjusted according to each second window determines the prediction result of third Window Index respectively.
S240 determines the accuracy rate of each prediction result according to the historical time sequence, and selects accuracy rate highest
Prediction result is the final prediction result of the third Window Index.
The mode of land parcel change trace may be used, calculate the historical forecast accuracy rate under the second length of window.Based on this, the standard
The calculating of true rate can be achieved by the steps of：
The comparison time series pair with the most similar specified quantity of each base time series is determined according to historical time sequence
The history index answered, the history index are that the adjacent index of the terminal of time series, the history index may be considered compared with
The corresponding true value of comparison time series；
According to the corresponding history index of each second window and prediction result, the accuracy rate of each prediction result is determined.
For example, predictablity rate can be indicated with 1-abs ((prediction result-true value)/true value), which indicates absolute value
Function.
By this specification embodiment, benchmark is used as by different length of window, future services index is predicted,
And historical data is combined to calculate the corresponding predictablity rate of each length of window, select the highest window of accuracy rate corresponding with this
Prediction result reduces influence of the benchmark length of window to prediction result with this, to reduce to benchmark window and historical time sequence
The requirement of the trend consistency of row, improves the accuracy of prediction result.
This programme use vector method it is not high to the coherence request of time trend, as recent traffic fluctuation it is larger or
Rise/decline is apparent, then being adapted to property of this programme adjusts predicted value；Secondly, this programme is high for the utilization rate of historical data,
Comparing vectorial selection is selected based on full dose historical data, and can utilize has nearest similar development in full dose historical data
Trend.
Wherein, the relationship between time series can be by between vector that the corresponding continuous numerical value of time series is constituted
Relationship indicate that be based on this, as shown in figure 4, in example in this specification embodiment, this method includes following step
Suddenly.
Step 401, source data is obtained.The data source refers to the data source of historical time sequence, which is the first window
The historical data of service operation in mouthful.
Step 402, source data is cleaned and is handled, generate historical time sequence.Specifically, according to certain rule
Data cleansing (such as carrying out noise reduction process to source data) is carried out to source data, in sequence generated time sequence indicator
Value；
Step 403, threshold value is calculated according to historical time sequence.For example, historical time sequence can be utilized, with case collimation method etc.
Method substantially determines the upper limit and lower limit of predicted value.The threshold value can be adjusted according to actual needs.
Step 404, vectorial (or second window) length is set with preset rules.For example, being adjusted by way of cycle
The length of vector.Wherein, vector length here refers to base vector and the vectorial length of comparison.If base vector is by the second window
The base time series that length is 3 corresponds to index composition, then vector length is 3.
Step 405, base vector is determined according to historical time sequence.Specifically, long at given vectorial (or second window)
After degree, last continuous several indexs of historical time sequence are taken to constitute basal orientation generally according to vector (or second window) length
Amount.
Step 406, it is selected according to historical time sequence and vectorial (or second window) length more vectorial.Specifically,
Time series compared with selection is identical with base vector length from historical time sequence generally in a manner of traversal, the comparison time
The corresponding index of sequence constitutes more vectorial.
Step 407, calculate and store base vector and the vectorial similarity of each comparison.For example, with the cosine angle of vector
Standard is portrayed as similarity (similarity is generally between 0-1).It is similar in addition to vector angle mentioned above to calculate vector
Cosine value can also be described by the Euclidean distance of vector and vectorial absolute value of the difference etc. movement.
For example, execution following steps can be recycled, until it is vectorial to traverse all comparisons：
Select the comparison not being traversed vectorial, the similarity for then calculating and storing it with base vector；
Judge whether to traverse all comparisons vectorial.
Step 408, selected the comparison of most like specified quantity vectorial according to similarity.For example, can be by the phase of storage
It is ranked up from high to low like degree, base vector and more vectorial more similar is indicated closer to 1 more than similarity, can selected and basal orientation
It is vectorial to measure the most similar preceding 20 comparisons.
Step 409, it determines prediction length, that is, needs the number for the index predicted, namely determine the length of third window.Such as
Time series numerical value is to need to predict two days weekends as unit of day, then prediction length is 2, and so on.
Step 410, according to the comparison selected is vectorial and prediction length calculates each more vectorial corresponding predicted value.Example
Such as, can and prediction length vectorial according to the comparison selected first extrapolate to obtain each comparison it is vectorial after actually occur value
(number is prediction length value), then proportion of utilization method be adjusted to actually occurring value and obtain predicted value, wherein predicted value
=value * Avg (base vector/more vectorial) are actually occurred, Avg indicates mean function.
Step 411, the exceptional value in the multiple predicted values being calculated is rejected.For example, using case collimation method rejecting abnormalities value.
Step 412, using the predicted value after adjusting thresholds rejecting abnormalities value.Specifically, using calculating the upper of threshold value
Limit and lower limit, to be adjusted to the predicted value after the rejecting of case collimation method, the numerical value higher than the upper limit will be pulled down to the upper limit, low
It will be raised to lower limit in the numerical value of lower limit.
Step 413, it is calculated as according to the reference vector of the length according to the predictor calculation average value after adjustment
Prediction result.
Step 414, the accuracy rate of prediction result is calculated according to historical time sequence.For example, to the mode of same land parcel change trace,
The historical forecast accuracy rate under this vector length is calculated, wherein predictablity rate can use 1-abs ((predicted value-true value)/true value)
It portrays, abs indicates ABS function.
Step 415, the accurate of the different corresponding prediction results of length base vector is can get by adjusting vector length angle value
Rate selects the highest prediction result of accuracy rate as final result.
This specification embodiment finds in history and now most close situation in such a way that vector is similar, and according to going through
Value is actually occurred to deduce following occurrence value after in history, and is further reduced in such a way that various optimizations are filtered different
The influence to predicted value such as constant value, to ensure the reliability of prediction result.
Corresponding to above method embodiment, this specification embodiment also provides a kind of index prediction meanss, referring to Fig. 5 institutes
Show, which may include：
Determination unit 601, in the historical time sequence of first window determine respectively in multiple second windows
Each isometric multiple relatively time serieses, each second window correspond to a base time series, and each second window corresponds to not
The length of window of same length of window, the first window is more than second window, the first window and second window
The terminal of mouth is identical；
First selecting unit 602, in the multiple relatively time series, selecting and each base time series respectively
The comparison time series of most similar specified quantity；
Predicting unit 603, for according to the comparison time sequence with the most similar specified quantity of each base time series
Row determine that the prediction result of third Window Index, the starting point of the third window are adjacent with the terminal of the second window respectively；
Second selecting unit 604 for determining the accuracy rate of each prediction result according to the historical time sequence, and is selected
Select the final prediction result that the highest prediction result of accuracy rate is the third Window Index.
In one example, the first selecting unit 602 is specifically used for：
A base vector is determined respectively according to each base time series, and each base vector is by its corresponding Ji Shijianxuliebao
The index included is constituted；
Determine that a comparison is vectorial according to each relatively time series, it is each more vectorial by its corresponding relatively time sequence
The index that row include is constituted；
According to base vector and more vectorial similarity, determine comparison with the most similar specified quantity of each base vector to
Amount, the comparison of the most similar specified quantity of each base vector it is vectorial by with the most similar specified quantity of each base time series
Comparison time series constitute.
In another example, the predicting unit 603 is specifically used for：
Specified number is determined respectively according to the comparison time series with the most similar specified quantity of each base time series
The prediction index of amount third Window Index；
Abnormal index in the corresponding specified quantity prediction index of each second window is rejected；
Determine the prediction result of third Window Index respectively according to the prediction index after rejecting.
In another example, the predicting unit 603 is specifically used for：
Determine the adjacent multiple continuous history identical with the third length of window of terminal of each relatively time series
Index；
Proportion of utilization method is adjusted each history index, respectively obtains the corresponding specified quantity of each second window
The prediction index of third Window Index；
In another example, described device further includes：
Threshold value determination unit, the threshold value for determining prediction index according to the historical time sequence.
In another example, the predicting unit 603 is specifically used for：
The prediction index for exceeding the threshold value in prediction index after rejecting is adjusted in the threshold value；
Prediction index after being adjusted according to each second window determines the prediction result of third Window Index respectively.
In another example, second selecting unit 604 is specifically used for：
The comparison time sequence with the most similar specified quantity of each base time series is determined according to the historical time sequence
Arrange corresponding history index；
According to the corresponding history index of each second window and prediction result, the accuracy rate of each prediction result is determined.
This specification embodiment also provides a kind of computer equipment, includes at least memory, processor and is stored in
On reservoir and the computer program that can run on a processor, the computer equipment can be implemented as the shape of index predictive server
Formula.Wherein, index prediction technique above-mentioned is realized when processor executes described program.This method includes at least：
Multiple ratios isometric with each of multiple second windows respectively are determined in the historical time sequence of first window
Compared with time series, each second window corresponds to a base time series, and each second window corresponds to different length of window, described
The length of window of first window is more than second window, and the first window is identical as the terminal of the second window；
The multiple relatively in time series, the ratio of selection and the most similar specified quantity of each base time series respectively
Compared with time series；
Third window is determined respectively according to the comparison time series with the most similar specified quantity of each base time series
The prediction result of mouth index, the starting point of the third window are adjacent with the terminal of the second window；
The accuracy rate of each prediction result is determined according to the historical time sequence, and selects the highest prediction knot of accuracy rate
Fruit is the final prediction result of the third Window Index.
In one example, described in the multiple relatively time series, it selects respectively with each base time series most
The comparison time series of similar specified quantity includes：
A base vector is determined respectively according to each base time series, and each base vector is by its corresponding Ji Shijianxuliebao
The index included is constituted；
Determine that a comparison is vectorial according to each relatively time series, it is each more vectorial by its corresponding relatively time sequence
The index that row include is constituted；
According to base vector and more vectorial similarity, determine comparison with the most similar specified quantity of each base vector to
Amount, the comparison of the most similar specified quantity of each base vector it is vectorial by with the most similar specified quantity of each base time series
Comparison time series constitute.
In another example, the comparison time described in the basis with the most similar specified quantity of each base time series
Sequence determines that the prediction result of third Window Index includes respectively：
Specified number is determined respectively according to the comparison time series with the most similar specified quantity of each base time series
The prediction index of amount third Window Index；
Abnormal index in the corresponding specified quantity prediction index of each second window is rejected；
Determine the prediction result of third Window Index respectively according to the prediction index after rejecting.
In another example, the comparison time described in the basis with the most similar specified quantity of each base time series
Sequence determines that the specified quantity prediction index of third Window Index includes respectively：
Determine the adjacent multiple continuous history identical with the third length of window of terminal of each relatively time series
Index；
Proportion of utilization method is adjusted each history index, respectively obtains the corresponding specified quantity of each second window
The prediction index of third Window Index；
In another example, the method further includes：
The threshold value of prediction index is determined according to the historical time sequence.
In another example, the prediction index according to after rejecting determines the prediction result of third Window Index respectively
Including：
The prediction index for exceeding the threshold value in prediction index after rejecting is adjusted in the threshold value；
Prediction index after being adjusted according to each second window determines the prediction result of third Window Index respectively.
In another example, described to determine that the accuracy rate of each prediction result includes according to the historical time sequence：
The comparison time sequence with the most similar specified quantity of each base time series is determined according to the historical time sequence
Arrange corresponding history index；
According to the corresponding history index of each second window and prediction result, the accuracy rate of each prediction result is determined.
Fig. 6 shows a kind of more specifically computer equipment structural schematic diagram that this specification embodiment is provided, should
Computer equipment may include：Processor 610, memory 620, input/output interface 630, communication interface 640 and bus 650.
Wherein processor 610, memory 620, input/output interface 630 and communication interface 640 pass through between the realization of bus 650
Communication connection inside equipment.
General CPU (Central Processing Unit, central processing unit), microprocessor may be used in processor 610
Device, application specific integrated circuit (Application Specific Integrated Circuit, ASIC) or one or
The modes such as multiple integrated circuits are realized, for executing relative program, to realize technical solution that this specification embodiment is provided.
ROM (Read Only Memory, read-only memory), RAM (Random Access may be used in memory 620
Memory, random access memory), static storage device, the forms such as dynamic memory realize.Memory 620 can store
Operating system and other applications are realizing technical solution that this specification embodiment is provided by software or firmware
When, relevant program code is stored in memory 620, and is executed by processor 610 to call.
Input/output interface 630 is for connecting input/output module, to realize information input and output.Input and output/
Module can be used as component Configuration (not shown) in a device, can also be external in equipment to provide corresponding function.Wherein
Input equipment may include keyboard, mouse, touch screen, microphone, various kinds of sensors etc., output equipment may include display,
Loud speaker, vibrator, indicator light etc..
Communication interface 640 is used for connection communication module (not shown), to realize the communication of this equipment and other equipment
Interaction.Wherein communication module can be realized by wired mode (such as USB, cable etc.) and be communicated, can also be wirelessly
(such as mobile network, WIFI, bluetooth etc.) realizes communication.
Bus 650 includes an access, in various components (such as processor 610, memory 620, the input/output of equipment
Interface 630 and communication interface 640) between transmit information.
It should be noted that although above equipment illustrates only processor 610, memory 620, input/output interface
630, communication interface 640 and bus 650, but in specific implementation process, which can also include realizing normal operation
Necessary other assemblies.In addition, it will be appreciated by those skilled in the art that, can also only include to realize in above equipment
Component necessary to this specification example scheme, without including all components shown in figure.
This specification embodiment also provides a kind of computer readable storage medium, is stored thereon with computer program, the journey
Index prediction technique above-mentioned is realized when sequence is executed by processor.This method includes at least：
Multiple ratios isometric with each of multiple second windows respectively are determined in the historical time sequence of first window
Compared with time series, each second window corresponds to a base time series, and each second window corresponds to different length of window, described
The length of window of first window is more than second window, and the first window is identical as the terminal of the second window；
The multiple relatively in time series, the ratio of selection and the most similar specified quantity of each base time series respectively
Compared with time series；
Third window is determined respectively according to the comparison time series with the most similar specified quantity of each base time series
The prediction result of mouth index, the starting point of the third window are adjacent with the terminal of the second window；
The accuracy rate of each prediction result is determined according to the historical time sequence, and selects the highest prediction knot of accuracy rate
Fruit is the final prediction result of the third Window Index.
In one example, described in the multiple relatively time series, it selects respectively with each base time series most
The comparison time series of similar specified quantity includes：
A base vector is determined respectively according to each base time series, and each base vector is by its corresponding Ji Shijianxuliebao
The index included is constituted；
Determine that a comparison is vectorial according to each relatively time series, it is each more vectorial by its corresponding relatively time sequence
The index that row include is constituted；
According to base vector and more vectorial similarity, determine comparison with the most similar specified quantity of each base vector to
Amount, the comparison of the most similar specified quantity of each base vector it is vectorial by with the most similar specified quantity of each base time series
Comparison time series constitute.
In another example, the comparison time described in the basis with the most similar specified quantity of each base time series
Sequence determines that the prediction result of third Window Index includes respectively：
Specified number is determined respectively according to the comparison time series with the most similar specified quantity of each base time series
The prediction index of amount third Window Index；
Abnormal index in the corresponding specified quantity prediction index of each second window is rejected；
Determine the prediction result of third Window Index respectively according to the prediction index after rejecting.
In another example, the comparison time described in the basis with the most similar specified quantity of each base time series
Sequence determines that the specified quantity prediction index of third Window Index includes respectively：
Determine the adjacent multiple continuous history identical with the third length of window of terminal of each relatively time series
Index；
Proportion of utilization method is adjusted each history index, respectively obtains the corresponding specified quantity of each second window
The prediction index of third Window Index；
In another example, the method further includes：
The threshold value of prediction index is determined according to the historical time sequence.
In another example, the prediction index according to after rejecting determines the prediction result of third Window Index respectively
Including：
The prediction index for exceeding the threshold value in prediction index after rejecting is adjusted in the threshold value；
Prediction index after being adjusted according to each second window determines the prediction result of third Window Index respectively.
In another example, described to determine that the accuracy rate of each prediction result includes according to the historical time sequence：
The comparison time sequence with the most similar specified quantity of each base time series is determined according to the historical time sequence
Arrange corresponding history index；
According to the corresponding history index of each second window and prediction result, the accuracy rate of each prediction result is determined.
Computer-readable medium includes permanent and non-permanent, removable and non-removable media can be by any method
Or technology realizes information storage.Information can be computer-readable instruction, data structure, the module of program or other data.
The example of the storage medium of computer includes, but are not limited to phase transition internal memory (PRAM), static RAM (SRAM), moves
State random access memory (DRAM), other kinds of random access memory (RAM), read-only memory (ROM), electric erasable
Programmable read only memory (EEPROM), fast flash memory bank or other memory techniques, read-only disc read only memory (CD-ROM) (CD-ROM),
Digital versatile disc (DVD) or other optical storages, magnetic tape cassette, tape magnetic disk storage or other magnetic storage apparatus
Or any other non-transmission medium, it can be used for storage and can be accessed by a computing device information.As defined in this article, it calculates
Machine readable medium does not include temporary computer readable media (transitory media), such as data-signal and carrier wave of modulation.
As seen through the above description of the embodiments, those skilled in the art can be understood that this specification
Embodiment can add the mode of required general hardware platform to realize by software.Based on this understanding, this specification is implemented
Substantially the part that contributes to existing technology can be expressed in the form of software products the technical solution of example in other words,
The computer software product can be stored in a storage medium, such as ROM/RAM, magnetic disc, CD, including some instructions are making
It is each to obtain computer equipment (can be personal computer, server or the network equipment etc.) execution this specification embodiment
Method described in certain parts of a embodiment or embodiment.
System, device, module or the unit that above-described embodiment illustrates can specifically realize by computer chip or entity,
Or it is realized by the product with certain function.A kind of typically to realize that equipment is computer, the concrete form of computer can
To be personal computer, laptop computer, cellular phone, camera phone, smart phone, personal digital assistant, media play
In device, navigation equipment, E-mail receiver/send equipment, game console, tablet computer, wearable device or these equipment
The combination of arbitrary several equipment.
Each embodiment in this specification is described in a progressive manner, identical similar portion between each embodiment
Point just to refer each other, and each embodiment focuses on the differences from other embodiments.Especially for device reality
For applying example, since it is substantially similar to the method embodiment, so describing fairly simple, related place is referring to embodiment of the method
Part explanation.The apparatus embodiments described above are merely exemplary, wherein described be used as separating component explanation
Module may or may not be physically separated, can be each module when implementing this specification example scheme
Function realize in the same or multiple software and or hardware.Can also select according to the actual needs part therein or
Person's whole module achieves the purpose of the solution of this embodiment.Those of ordinary skill in the art are not the case where making the creative labor
Under, you can to understand and implement.
The above is only the specific implementation mode of this specification embodiment, it is noted that for the general of the art
For logical technical staff, under the premise of not departing from this specification embodiment principle, several improvements and modifications can also be made, this
A little improvements and modifications also should be regarded as the protection domain of this specification embodiment.
Claims (15)
1. a kind of index prediction technique, which is characterized in that the method includes：
When being determined in the historical time sequence of first window respectively compared with isometric multiple of each of multiple second windows
Between sequence, each second window corresponds to a base time series, and each second window corresponds to different length of window, described first
The length of window of window is more than second window, and the first window is identical as the terminal of the second window；
In the multiple relatively time series, when selecting the comparison with the most similar specified quantity of each base time series respectively
Between sequence；
Determine that third window refers to respectively according to the comparison time series with the most similar specified quantity of each base time series
Target prediction result, the starting point of the third window are adjacent with the terminal of the second window；
Determine the accuracy rate of each prediction result according to the historical time sequence, and select the highest prediction result of accuracy rate for
The final prediction result of the third Window Index.
2. according to the method described in claim 1, it is characterized in that, described relatively in time series, select respectively the multiple
It selects and includes with the comparison time series of the most similar specified quantity of each base time series：
Determine that a base vector, each base vector include by its corresponding base time series respectively according to each base time series
Index is constituted；
Determine that a comparison is vectorial according to each relatively time series, it is each more vectorial by its corresponding relatively time series packet
The index included is constituted；
According to base vector and more vectorial similarity, determination is vectorial with the comparison of the most similar specified quantity of each base vector,
The comparison of the most similar specified quantity of each base vector it is vectorial by with the most similar specified quantity of each base time series
Compare time series composition.
3. according to the method described in claim 1, it is characterized in that, most similar with each base time series described in the basis
The comparison time series of specified quantity determines that the prediction result of third Window Index includes respectively：
Specified quantity is determined respectively according to the comparison time series with the most similar specified quantity of each base time series
The prediction index of third Window Index；
Abnormal index in the corresponding specified quantity prediction index of each second window is rejected；
Determine the prediction result of third Window Index respectively according to the prediction index after rejecting.
4. according to the method described in claim 3, it is characterized in that, most similar with each base time series described in the basis
The comparison time series of specified quantity determines that the specified quantity prediction index of third Window Index includes respectively：
Determine the adjacent multiple continuous history indexs identical with the third length of window of terminal of each relatively time series；
Proportion of utilization method is adjusted each history index, respectively obtains the corresponding specified quantity third of each second window
The prediction index of Window Index.
5. according to the method described in any of claim 1 to 4, which is characterized in that the method further includes：
The threshold value of prediction index is determined according to the historical time sequence.
6. according to the method described in claim 5, it is characterized in that, the prediction index according to after rejecting determines third respectively
The prediction result of Window Index includes：
The prediction index for exceeding the threshold value in prediction index after rejecting is adjusted in the threshold value；
Prediction index after being adjusted according to each second window determines the prediction result of third Window Index respectively.
7. according to the method described in any of claim 1 to 4, which is characterized in that described true according to the historical time sequence
The accuracy rate of each prediction result includes calmly：
The comparison time series pair with the most similar specified quantity of each base time series is determined according to the historical time sequence
The history index answered；
According to the corresponding history index of each second window and prediction result, the accuracy rate of each prediction result is determined.
8. a kind of index prediction meanss, which is characterized in that described device includes：
Determination unit, for determining isometric with each of multiple second windows respectively in the historical time sequence of first window
Multiple relatively time serieses, each second window corresponds to a base time series, and each second window corresponds to different windows
The length of window of length, the first window is more than second window, the terminal of the first window and second window
It is identical；
First selecting unit, in the multiple relatively time series, selecting respectively most close with each base time series
Specified quantity comparison time series；
Predicting unit, for according to described true with the comparison time series difference of the most similar specified quantity of each base time series
Determine the prediction result of third Window Index, the starting point of the third window is adjacent with the terminal of the second window；
Second selecting unit, the accuracy rate for determining each prediction result according to the historical time sequence, and select accurately
The highest prediction result of rate is the final prediction result of the third Window Index.
9. device according to claim 8, which is characterized in that the first selecting unit is specifically used for：
Determine that a base vector, each base vector include by its corresponding base time series respectively according to each base time series
Index is constituted；
Determine that a comparison is vectorial according to each relatively time series, it is each more vectorial by its corresponding relatively time series packet
The index included is constituted；
According to base vector and more vectorial similarity, determination is vectorial with the comparison of the most similar specified quantity of each base vector,
The comparison of the most similar specified quantity of each base vector it is vectorial by with the most similar specified quantity of each base time series
Compare time series composition.
10. device according to claim 8, which is characterized in that the predicting unit is specifically used for：
Specified quantity is determined respectively according to the comparison time series with the most similar specified quantity of each base time series
The prediction index of third Window Index；
Abnormal index in the corresponding specified quantity prediction index of each second window is rejected；
Determine the prediction result of third Window Index respectively according to the prediction index after rejecting.
11. device according to claim 10, which is characterized in that the predicting unit is specifically used for：
Determine the adjacent multiple continuous history indexs identical with the third length of window of terminal of each relatively time series；
Proportion of utilization method is adjusted each history index, respectively obtains the corresponding specified quantity third of each second window
The prediction index of Window Index.
12. according to the device described in claim 8-11 any one, which is characterized in that described device further includes：
Threshold value determination unit, the threshold value for determining prediction index according to the historical time sequence.
13. device according to claim 12, which is characterized in that the predicting unit is specifically used for：
The prediction index for exceeding the threshold value in prediction index after rejecting is adjusted in the threshold value；
Prediction index after being adjusted according to each second window determines the prediction result of third Window Index respectively.
14. according to the device described in claim 8-11 any one, which is characterized in that second selecting unit is specifically used
In：
The comparison time series pair with the most similar specified quantity of each base time series is determined according to the historical time sequence
The history index answered；
According to the corresponding history index of each second window and prediction result, the accuracy rate of each prediction result is determined.
15. a kind of computer equipment, including memory, processor and storage are on a memory and the meter that can run on a processor
Calculation machine program, which is characterized in that the processor realizes following steps when executing described program：
Sampled data when acquisition system normal operation, the normal sample during the sampled data is gathered as training；
According to prefabricated Rule abnormal data, cycle executes following steps, until the recognition effect of index prediction model reaches
It is expected that carrying out index prediction to data to be tested to use recognition effect to reach expected index prediction model：
The abnormal data is extended, the abnormal data and the abnormal data of extension are increased as exceptional sample in institute
It states in training set；
The index prediction model is trained according to the training set after increase abnormal data, and determines the index prediction
The recognition effect of model；
When the recognition effect of the index prediction model is less than being expected, according to the new abnormal data of the prefabricated Rule.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810483404.9A CN108665113A (en) | 2018-05-18 | 2018-05-18 | Index prediction technique and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810483404.9A CN108665113A (en) | 2018-05-18 | 2018-05-18 | Index prediction technique and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108665113A true CN108665113A (en) | 2018-10-16 |
Family
ID=63777153
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810483404.9A Pending CN108665113A (en) | 2018-05-18 | 2018-05-18 | Index prediction technique and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108665113A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111125195A (en) * | 2019-12-25 | 2020-05-08 | 亚信科技（中国）有限公司 | Data anomaly detection method and device |
| WO2020185155A1 (en) * | 2019-03-13 | 2020-09-17 | Rublix Development Pte. Ltd. | A system and method for crediting a predictive entity |
-
2018
- 2018-05-18 CN CN201810483404.9A patent/CN108665113A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020185155A1 (en) * | 2019-03-13 | 2020-09-17 | Rublix Development Pte. Ltd. | A system and method for crediting a predictive entity |
| CN111125195A (en) * | 2019-12-25 | 2020-05-08 | 亚信科技（中国）有限公司 | Data anomaly detection method and device |
| CN111125195B (en) * | 2019-12-25 | 2023-09-08 | 亚信科技（中国）有限公司 | Data anomaly detection method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| TW202008237A (en) | Method and device for training prediction model for new scenario | |
| CA3059744C (en) | Strategy searching in strategic interaction between parties | |
| KR20170134601A (en) | Data processing method and apparatus | |
| JP6907664B2 (en) | Methods and equipment used to predict non-stationary time series data | |
| Isah et al. | The hidden predictive power of cryptocurrencies and QE: Evidence from US stock market | |
| Chehrazi et al. | Dynamic valuation of delinquent credit-card accounts | |
| Ritschl et al. | Currency versus banking in the financial crisis of 1931 | |
| CN106610818A (en) | Attribute determining method and apparatus | |
| AU2021225802A1 (en) | Machine-learning techniques to generate recommendations for risk mitigation | |
| CN108537460A (en) | Consumer's risk prediction technique and system | |
| CN105844516A (en) | Method and device for evaluating enterprise credit | |
| CN110533531A (en) | Identify method and apparatus, medium and the electronic equipment of abnormal account | |
| Plakandaras et al. | The informational content of the term spread in forecasting the us inflation rate: a nonlinear approach | |
| CN110020865A (en) | Transaction identification method and device | |
| CN108665113A (en) | Index prediction technique and device | |
| CN109146148A (en) | A kind of determination method and device of remaining sum prediction deviation reason | |
| CN109102282A (en) | A kind of Multiple Currencies reimbursement method for processing business and device | |
| JP5863137B1 (en) | Inheritance tax countermeasure support system | |
| Choi et al. | Regime Shifts in Price‐Dividend Ratios and Expected Stock Returns: A Present‐Value Approach | |
| CN110232620A (en) | Trade company's label determines method, apparatus, electronic equipment and readable storage medium storing program for executing | |
| JP2017084229A (en) | Investment simulation device and method | |
| JP2020013229A (en) | Device, method and program for calculating default probability | |
| CN113610627A (en) | Data processing method and device for risk early warning | |
| CN106598987A (en) | Information recommendation method and device | |
| Martınez Jaramillo et al. | Systemic risk analysis by means of network theory: an empirical study of the Mexican banking system |
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 | ||
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20201022 Address after: Cayman Enterprise Centre, 27 Hospital Road, George Town, Grand Cayman Islands Applicant after: Innovative advanced technology Co.,Ltd. Address before: Cayman Enterprise Centre, 27 Hospital Road, George Town, Grand Cayman Islands Applicant before: Advanced innovation technology Co.,Ltd. Effective date of registration: 20201022 Address after: Cayman Enterprise Centre, 27 Hospital Road, George Town, Grand Cayman Islands Applicant after: Advanced innovation technology Co.,Ltd. Address before: A four-storey 847 mailbox in Grand Cayman Capital Building, British Cayman Islands Applicant before: Alibaba Group Holding Ltd. |
|
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181016 |