CN116883060A - Target prediction method, storage medium and device based on time sequence associated data - Google Patents

Target prediction method, storage medium and device based on time sequence associated data Download PDF

Info

Publication number
CN116883060A
CN116883060A CN202311147980.3A CN202311147980A CN116883060A CN 116883060 A CN116883060 A CN 116883060A CN 202311147980 A CN202311147980 A CN 202311147980A CN 116883060 A CN116883060 A CN 116883060A
Authority
CN
China
Prior art keywords
attribute
dimension
target
value
attribute dimension
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
Application number
CN202311147980.3A
Other languages
Chinese (zh)
Inventor
陈方博
房馨
龚腾飞
张恩瑞
张宏亮
潘欣
刘鹏
白蒙
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Ccb Housing Service Co ltd
Original Assignee
Ccb Housing Service Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Ccb Housing Service Co ltd filed Critical Ccb Housing Service Co ltd
Priority to CN202311147980.3A priority Critical patent/CN116883060A/en
Publication of CN116883060A publication Critical patent/CN116883060A/en
Pending legal-status Critical Current

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
    • 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
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9537—Spatial or temporal dependent retrieval, e.g. spatiotemporal queries
    • 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
    • G06Q30/0202—Market predictions or forecasting for commercial activities

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Accounting & Taxation (AREA)
  • Development Economics (AREA)
  • Finance (AREA)
  • Strategic Management (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Physics & Mathematics (AREA)
  • Game Theory and Decision Science (AREA)
  • General Business, Economics & Management (AREA)
  • Marketing (AREA)
  • Economics (AREA)
  • General Engineering & Computer Science (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The present invention relates to the field of data processing, and in particular, to a target prediction method, a storage medium, and a device based on time-series associated data. Comprising the following steps: according to the target attribute dimension corresponding to the predicted target; acquiring an attribute value corresponding to a target acquisition period in each first attribute dimension according to a prediction time t corresponding to a prediction attribute value and a look-ahead period value between each first attribute dimension and a target attribute dimension, and generating initial feature data corresponding to a prediction target; and inputting the initial characteristic data into a target attribute value determination model to generate a prediction result corresponding to the prediction target. According to the method and the device, the attribute values generated by the corresponding target acquisition period in each first attribute dimension are processed according to the correlation and the advance period value, so that the attribute values generated by the corresponding target acquisition period in each first attribute dimension can be more accurately and comprehensively selected, and finally the reliability of the obtained prediction result is improved.

Description

Target prediction method, storage medium and device based on time sequence associated data
Technical Field
The present invention relates to the field of data processing, and in particular, to a target prediction method, a storage medium, and a device based on time-series associated data.
Background
Generally, according to the development rule of an object, the trend of the future development rule of an object is affected by a plurality of dimension factors related to the object. For example, the price of an article in an economic market is taken as an example, and the specific price of the article is basically influenced by the supply of the article by the market and the demand of the article by the market. Therefore, in order to more accurately predict the future development trend of a certain attribute value of an object, it is necessary to more accurately determine the attribute value of each factor affecting the attribute value. Since each influence factor can affect the target prediction result, the required time intervals are not the same, so the generation times of a plurality of influence factors that can affect the target prediction result are not the same. However, in the prior art, when predicting the target, the characteristics are not considered, so that the attribute dimension which can affect the target prediction result in the obtained initial feature data is not comprehensive, the time sequence is inaccurate, and the problem of low reliability of the finally obtained prediction result is caused.
Disclosure of Invention
Aiming at the technical problems, the invention adopts the following technical scheme:
according to an aspect of the present invention, there is provided a target prediction method based on time-series associated data, the method comprising the steps of:
and acquiring a plurality of first attribute dimensions from a preset feature library according to the target attribute dimensions corresponding to the predicted target. The first attribute dimension is an attribute dimension in the feature library having a correlation with the target attribute dimension greater than a first correlation threshold. The first attribute dimension comprises attribute values corresponding to a plurality of acquisition periods respectively.
And acquiring an attribute value corresponding to a target acquisition period in each first attribute dimension according to the prediction time t corresponding to the predicted attribute value and a look-ahead period value between each first attribute dimension and the target attribute dimension, and generating initial characteristic data corresponding to a predicted target. Wherein, the target acquisition period T corresponding to the a first attribute dimension b =t-f b 。f b Is the look-ahead period value between the b-th first attribute dimension and the target attribute dimension.
And inputting the initial characteristic data into a target attribute value determination model to generate a prediction result corresponding to the prediction target.
The value of the advance period between each first attribute dimension and the target attribute dimension is obtained according to the following steps:
historical feature data is obtained. The historical characteristic data comprises attribute values corresponding to each first attribute dimension and the target attribute dimension in a plurality of continuous historical acquisition periods.
And generating the interval correlation degree between the target attribute dimension and each first attribute dimension under the condition of different acquisition intervals according to the attribute value corresponding to the target attribute dimension and the attribute value corresponding to each first attribute dimension. Wherein, under the condition of spacing e acquisition periods, the interval correlation degree R of the target attribute dimension and the b first attribute dimension e b The following relationship is satisfied:
；
wherein M is a And acquiring attribute values corresponding to the period a in the dimension of the target attribute. Y is Y g And the characteristic mean value corresponding to the target attribute dimension. B (B) a-e And the attribute value corresponding to the a-e acquisition period in the b first attribute dimension is obtained. Y is Y b And the characteristic mean value corresponding to the b first attribute dimension. z is the firstThe number of attribute values contained in an attribute dimension. e=0, 1, … x, x being the upper interval limit, x<z。
And taking the collection period interval number corresponding to the correlation extremum of the target attribute dimension and each first attribute dimension as a preceding period value between the target attribute dimension and each first attribute dimension. The correlation extremum is the maximum of the absolute value of the interval correlation between the target attribute dimension and the first attribute dimension.
According to a second aspect of the present invention, there is provided a non-transitory computer readable storage medium storing a computer program which when executed by a processor implements a target prediction method based on time-series related data as described above.
According to a third aspect of the present invention, there is provided an electronic device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing a method of target prediction based on time-series related data as described above when the computer program is executed by the processor.
The invention has at least the following beneficial effects:
according to the method, before the initial characteristic data is input into the target attribute value determining model, a plurality of first attribute dimensions with higher relativity are acquired from the preset characteristic library according to the target attribute dimensions corresponding to the predicted target, so that all factors influencing the target to be predicted can be determined more accurately and comprehensively, and the coverage of the initial characteristic data is improved. And then acquiring the attribute value obtained by the corresponding acquisition period from each first attribute dimension according to the prediction time t corresponding to the predicted attribute value and the advance period value between each first attribute dimension and the target attribute dimension. Through the advance period value corresponding to each first attribute dimension, it can be determined that each first attribute dimension can advance by a plurality of acquisition periods, and the influence on the target attribute dimension at the prediction time is generated. And then the attribute value generated by the corresponding target acquisition period in each first attribute dimension can be more accurately selected, and finally the attribute value of each attribute dimension which is more in line with the time sequence rule generates initial characteristic data, so that the reliability of the obtained prediction result is improved.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of a target prediction method based on time-series associated data according to an embodiment of the present invention.
Fig. 2 is a schematic diagram of correlation between price comparability indexes and partial index characteristic time differences of newly-built commercial residences in a certain city according to an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to fall within the scope of the invention.
According to an aspect of the present invention, as shown in fig. 1, there is provided a target prediction method based on time-series associated data, the method comprising the steps of:
s100: and acquiring a plurality of first attribute dimensions from a preset feature library according to the target attribute dimensions corresponding to the predicted target. The first attribute dimension is an attribute dimension in the feature library having a correlation with the target attribute dimension greater than a first correlation threshold. The first attribute dimension comprises attribute values corresponding to a plurality of acquisition periods respectively. The first correlation threshold may take a value in the interval of [0.4,1].
Specifically, S100 includes:
s101: and generating a correlation value of the target attribute dimension and each attribute dimension according to the target attribute dimension and a plurality of attribute values corresponding to each attribute dimension in the preset feature library.
Preferably, a covariance algorithm or a correlation algorithm is used for processing a plurality of attribute values corresponding to each attribute dimension in a target attribute dimension and a preset feature library, and a correlation value of the target attribute dimension and each attribute dimension is generated.
S102: and acquiring a first attribute dimension from the attribute dimensions according to the correlation value corresponding to each attribute dimension. The first attribute dimension is an attribute dimension having a correlation value greater than a first correlation threshold.
In the step, each attribute dimension having an influence on the target attribute dimension is more accurately and comprehensively determined mainly through the relevance value. In general, in different usage scenarios, the preset feature library will also be different, and the target attribute dimension is different from each corresponding attribute dimension. Taking real estate scenarios as an example for illustration:
the attribute dimensions may be summarized from the following land markets, property development and investment, commodity supply, commodity demand, property price, property finance, property related importance ratios, property related other directions, and the like, according to what may be involved in the property transaction. Specifically, the following attribute dimensions can be included:
land acquisition area, land price, commodity house selling, area for sale, commodity house approval pre (sales) condition, commodity house removal period, commodity house price, second-hand house sales area, newly built commodity house sales area, RMB actual effective exchange rate, medium-and-long term loan interest rate, legal deposit preparation rate, PPI (Producer Price Index, production price index) and the like.
S200: and acquiring an attribute value corresponding to a target acquisition period in each first attribute dimension according to the prediction time t corresponding to the predicted attribute value and a look-ahead period value between each first attribute dimension and the target attribute dimension, and generating initial characteristic data corresponding to a predicted target. Wherein, the target acquisition period T corresponding to the a first attribute dimension b =t-f b 。f b Is the look-ahead period value between the b-th first attribute dimension and the target attribute dimension.
Since the time at which each attribute dimension affects the target attribute dimension is different as shown in FIG. 2, according to T b =t-f b The attribute values in each first attribute dimension that can be used to make the prediction data can be more accurately determined. And finally, generating initial characteristic data by attribute values of each attribute dimension which better accords with a time sequence rule, thereby improving the reliability of the obtained prediction result.
Specifically, before S200, the method further includes:
s210: if the advance period value corresponding to the first attribute dimension is greater than zero, determining that the first attribute dimension is one data dimension in the initial feature data.
S220: if the advance period value corresponding to the first attribute dimension is smaller than zero, determining the first attribute dimension as one data dimension in the verification data.
Specifically, the advance cycle value in the present invention may be determined according to steps S201 to S203 described below.
Since the leading period value represents the corresponding first attribute dimension, the target to be predicted is affected a few acquisition periods in advance. So if the look-ahead period value is positive, it indicates that the first attribute dimension will change correspondingly before the predicted outcome occurs. If the look-ahead period value is negative, it indicates that the first attribute dimension will undergo a corresponding change after the predicted outcome occurs. Therefore, the first attribute dimension with a positive leading period value can be used as prediction data, and the first attribute dimension with a negative leading period value can be used as verification data. Based on the characteristics, the first attribute dimension with the positive value of the advanced period value can be used as training data in the training process of determining the model for the target attribute value, and the first attribute dimension with the negative value of the advanced period value can be used as verification data.
S300: and inputting the initial characteristic data into a target attribute value determination model to generate a prediction result corresponding to the prediction target.
The target attribute value determining model used in the embodiment may be a model obtained by training the training data of the corresponding scene through the existing neural network model.
The value of the advance period between each first attribute dimension and the target attribute dimension is obtained according to the following steps:
s201: historical feature data is obtained. The historical characteristic data comprises attribute values corresponding to each first attribute dimension and the target attribute dimension in a plurality of continuous historical acquisition periods.
Specifically, the historical data may be the historical data of the first two years, and the acquisition period may be one natural month, so that the historical characteristic data includes data corresponding to 24 months. And the data dimension is the same for each month. Taking real estate application scenario as an example, the historical data can be obtained from the statistics data of the network signature record data and the related statistics departments.
S201 includes:
s211: and acquiring an attribute value of each attribute dimension generated in a preset history period.
S221: the preset history period is divided into a plurality of acquisition periods. The acquisition period comprises at least one multi-value dimension, wherein the multi-value dimension is an attribute dimension for acquiring a plurality of attribute values in the corresponding acquisition period.
S231: and generating a target attribute value corresponding to each multi-value dimension according to the average value of the attribute values included in each multi-value dimension.
There are typically attribute dimensions that produce multiple attribute values within a month, such as sales unit price of a new good in an area, multiple changes within a month, or multiple new good cells in the area are sold at the same time. Therefore, the sales unit price of a plurality of newly built commodity houses can be generated in one month, and the average value of the sales unit price can be used for carrying out subsequent calculation processing in order to ensure the accuracy of data. The continuous data can be discretized through the processing, so that one attribute dimension corresponds to a more accurate attribute value.
S202: and generating the interval correlation degree between the target attribute dimension and each first attribute dimension under the condition of different acquisition intervals according to the attribute value corresponding to the target attribute dimension and the attribute value corresponding to each first attribute dimension. Wherein at intervals ofIn the case of e acquisition periods, the interval correlation degree R between the target attribute dimension and the b first attribute dimension e b The following relationship is satisfied:
；
wherein M is a And acquiring attribute values corresponding to the period a in the dimension of the target attribute. Y is Y g And the characteristic mean value corresponding to the target attribute dimension. B (B) a-e And the attribute value corresponding to the a-e acquisition period in the b first attribute dimension is obtained. Y is Y b And the characteristic mean value corresponding to the b first attribute dimension. z is the number of attribute values contained in the first attribute dimension. e=0, 1, … x, x being the upper interval limit, x<z。
The above formula is a formula for calculating the interval correlation degree between the target attribute dimension and each first attribute dimension. Wherein, the two attribute values of the calculated correlation are controlled by e to be different by a few acquisition periods. When e=1, the correlation between the attribute values of the target attribute dimension and the first attribute dimension, which are different by 1 acquisition period, such as the correlation between the attribute value of the 12 th month of the target attribute dimension and the attribute value of the 11 th month of the first attribute dimension, is calculated.
If e >0, it indicates that the first attribute dimension will change before the predicted outcome occurs.
If e <0, it indicates that the first attribute dimension will undergo a corresponding change after the occurrence of the predicted result
When R is e b <When 0, the target attribute dimension and each first attribute dimension show negative correlation characteristics, when R e b >And 0, the target attribute dimension and each first attribute dimension show positive correlation characteristics.
Generally, the interval correlation takes absolute value, 0-0.1 indicates no correlation, 0.1-0.3 is weak correlation, 0.3-0.5 is medium correlation, and 0.5-1.0 is strong correlation.
S203: and taking the collection period interval number corresponding to the correlation extremum of the target attribute dimension and each first attribute dimension as a preceding period value between the target attribute dimension and each first attribute dimension. The correlation extremum is the maximum of the absolute value of the interval correlation between the target attribute dimension and the first attribute dimension.
After the processing in S202 described above, a relationship diagram between the interval correlation degree of the target attribute dimension and each first attribute dimension and the advance period value may be generated, as shown in fig. 2 below. As can be seen from the figure, the interval correlation with the target attribute dimension is maximized when the advance period value is larger than that of each first attribute dimension. Thus, the advance period value between the first attribute dimensions can be more accurately determined. Illustrated in fig. 2:
the target attribute dimension is the newly built commodity residence price, wherein when the first attribute dimension (PPI) reaches the maximum positive correlation with the target attribute dimension when the advanced period value is-12, the advanced period value corresponding to the PPI is-12.
The first attribute dimension (medium-to-long term loan interest rate) is most inversely correlated with the target attribute dimension at a lead period value of 5, so the lead period value corresponding to the medium-to-long term loan interest rate is 5.
As another possible embodiment of the present invention, the L-BFGS algorithm is used to iteratively calculate the initial model during training of the target attribute value determination model.
In this embodiment, the initial model may be an existing neural network model, and then a required training data set and a verification data set are obtained from the historical data according to the actual usage scenario, and then the initial model is trained so as to have the prediction capability of response.
It should be noted that in the process of constructing the training data set, the correlation between attribute dimensions is considered, and when the target attribute dimension is affected, the corresponding attribute dimension has advance and retard in time sequence relative to the target attribute dimension. Specifically, the corresponding correlation and time sequence processing can be performed on the training data by using S100 and S200 in the previous embodiment, so that the finally obtained training data is more accurate and comprehensive, and the prediction accuracy of the obtained target attribute value determination model is further improved.
In addition, in some usage scenarios, the obtained training data includes a large number of attribute dimensions, so that a large number of parameters are generated in the iterative computation process of the model, taking a real estate model as an example, the iterative computation is performed by using a BFGS algorithm, each iterative computation needs a Hesse matrix obtained by the previous iteration, the storage space of the matrix is at least N (n+1)/2, N is a feature dimension, and for the real estate, the required storage space is very huge, and the space occupied by one iteration result is usually about 3-6G. Thus, a great deal of memory space is occupied during the training process.
The L-BFGS algorithm in this embodiment is a modified version of the BFGS (Broyden-Fletcher-Goldfarb-Shanno, pseudo-Newton method) algorithm (iterative algorithm for solving unconstrained optimization problems), using the most recent m iterations of information at a given positive integer mAnd->、/>，/>It is possible to deduce +.>. The large-scale problem and the high-dimensional problem can be better processed relative to the BFGS algorithm without manually selecting a step factor or a descending direction.
Specifically, the L-BFGS algorithm constructs a recursive formula for the last result as follows:
because ofAlso scalar, the iterative formula can be further written as:
based on the above, the conventional improvement method is thatThe above can be modified as:
then useRepresentation->The expression of (2) can be expressed as:
here let the orderThen->。
It can also be expressed as:
after this step, the L-BFGS algorithm sets the value of m,。
when k+1 is less than or equal to m:
when k+1 > m:
the L-BFGS algorithm is an improved version of the BFGS algorithm, using the most recent m iterations of information given a positive integer mAnd->、/>、/>It is possible to deduce +.>. The large-scale problem and the high-dimensional problem can be better processed relative to the BFGS algorithm without manually selecting a step factor or a descending direction. Wherein, the meaning of the parameters in the formula is the prior art and is not described herein.
By using the L-BFGS algorithm (limited-memory BFGS, quasi-newton algorithm) to perform iterative computation on the initial model, the basic idea of L-BFGS is to replace the Hesse matrix of the previous iteration by storing a small amount of data of the previous m iterations. Specifically, only two target vectors in m iterations and a diagonal matrix are needed to be stored, and a total of 2 x m+1N-dimensional vectors (in practical application, m generally takes a value between 4 and 7) are needed to be stored, so that the data to be stored is far smaller than the matrix obtained by the BFGS algorithm.
Furthermore, although the steps of the methods in the present disclosure are depicted in a particular order in the drawings, this does not require or imply that the steps must be performed in that particular order or that all illustrated steps be performed in order to achieve desirable results. Additionally or alternatively, certain steps may be omitted, multiple steps combined into one step to perform, and/or one step decomposed into multiple steps to perform, etc.
From the above description of embodiments, those skilled in the art will readily appreciate that the example embodiments described herein may be implemented in software, or may be implemented in software in combination with the necessary hardware. Thus, the technical solution according to the embodiments of the present disclosure may be embodied in the form of a software product, which may be stored in a non-volatile storage medium (may be a CD-ROM, a U-disk, a mobile hard disk, etc.) or on a network, including several instructions to cause a computing device (may be a personal computer, a server, a mobile terminal, or a network device, etc.) to perform the method according to the embodiments of the present disclosure.
In an exemplary embodiment of the present disclosure, an electronic device capable of implementing the above method is also provided.
Those skilled in the art will appreciate that the various aspects of the invention may be implemented as a system, method, or program product. Accordingly, aspects of the invention may be embodied in the following forms, namely: an entirely hardware embodiment, an entirely software embodiment (including firmware, micro-code, etc.) or an embodiment combining hardware and software aspects may be referred to herein as a "circuit," module "or" system.
An electronic device according to this embodiment of the invention. The electronic device is merely an example, and should not impose any limitations on the functionality and scope of use of embodiments of the present invention.
The electronic device is in the form of a general purpose computing device. Components of an electronic device may include, but are not limited to: the at least one processor, the at least one memory, and a bus connecting the various system components, including the memory and the processor.
Wherein the memory stores program code that is executable by the processor to cause the processor to perform steps according to various exemplary embodiments of the present invention described in the above section of the exemplary method of this specification.
The storage may include readable media in the form of volatile storage, such as Random Access Memory (RAM) and/or cache memory, and may further include Read Only Memory (ROM).
The storage may also include a program/utility having a set (at least one) of program modules including, but not limited to: an operating system, one or more application programs, other program modules, and program data, each or some combination of which may include an implementation of a network environment.
The bus may be one or more of several types of bus structures including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor, or a local bus using any of a variety of bus architectures.
The electronic device may also communicate with one or more external devices (e.g., keyboard, pointing device, bluetooth device, etc.), with one or more devices that enable a user to interact with the electronic device, and/or with any device (e.g., router, modem, etc.) that enables the electronic device to communicate with one or more other computing devices. Such communication may be through an input/output (I/O) interface. And, the electronic device may also communicate with one or more networks such as a Local Area Network (LAN), a Wide Area Network (WAN), and/or a public network, such as the Internet, through a network adapter. The network adapter communicates with other modules of the electronic device via a bus. It should be appreciated that although not shown, other hardware and/or software modules may be used in connection with an electronic device, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID systems, tape drives, data backup storage systems, and the like.
From the above description of embodiments, those skilled in the art will readily appreciate that the example embodiments described herein may be implemented in software, or may be implemented in software in combination with the necessary hardware. Thus, the technical solution according to the embodiments of the present disclosure may be embodied in the form of a software product, which may be stored in a non-volatile storage medium (may be a CD-ROM, a U-disk, a mobile hard disk, etc.) or on a network, including several instructions to cause a computing device (may be a personal computer, a server, a terminal device, or a network device, etc.) to perform the method according to the embodiments of the present disclosure.
In an exemplary embodiment of the present disclosure, a computer-readable storage medium having stored thereon a program product capable of implementing the method described above in the present specification is also provided. In some possible embodiments, the aspects of the invention may also be implemented in the form of a program product comprising program code for causing a terminal device to carry out the steps according to the various exemplary embodiments of the invention as described in the "exemplary method" section of this specification, when the program product is run on the terminal device.
The program product may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium can be, for example, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium would include the following: an electrical connection having one or more wires, a portable disk, a hard disk, random Access Memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
The computer readable signal medium may include a data signal propagated in baseband or as part of a carrier wave with readable program code embodied therein. Such a propagated data signal may take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination of the foregoing. A readable signal medium may also be any readable medium that is not a readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.
Program code embodied on a readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.
Program code for carrying out operations of the present invention may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C++ or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device, partly on a remote computing device, or entirely on the remote computing device or server. In the case of remote computing devices, the remote computing device may be connected to the user computing device through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computing device (e.g., connected via the Internet using an Internet service provider).
Furthermore, the above-described drawings are only schematic illustrations of processes included in the method according to the exemplary embodiment of the present invention, and are not intended to be limiting. It will be readily appreciated that the processes shown in the above figures do not indicate or limit the temporal order of these processes. In addition, it is also readily understood that these processes may be performed synchronously or asynchronously, for example, among a plurality of modules.
It should be noted that although in the above detailed description several modules or units of a device for action execution are mentioned, such a division is not mandatory. Indeed, the features and functionality of two or more modules or units described above may be embodied in one module or unit in accordance with embodiments of the present disclosure. Conversely, the features and functions of one module or unit described above may be further divided into a plurality of modules or units to be embodied.
The present invention is not limited to the above embodiments, and any changes or substitutions that can be easily understood by those skilled in the art within the technical scope of the present invention are intended to be included in the scope of the present invention. Therefore, the protection scope of the invention is subject to the protection scope of the claims.

Claims (10)

1. A method of target prediction based on time-series associated data, the method comprising the steps of:
acquiring a plurality of first attribute dimensions from a preset feature library according to target attribute dimensions corresponding to a predicted target; the first attribute dimension is an attribute dimension of which the correlation with the target attribute dimension in the feature library is larger than a first correlation threshold; the first attribute dimension comprises attribute values corresponding to a plurality of acquisition periods respectively;
acquiring an attribute value corresponding to a target acquisition period in each first attribute dimension according to a prediction time t corresponding to a prediction attribute value and a look-ahead period value between each first attribute dimension and a target attribute dimension, and generating initial feature data corresponding to the prediction target; wherein, the target acquisition period T corresponding to the a first attribute dimension b =t-f b ；f b A look-ahead period value between the b-th first attribute dimension and the target attribute dimension;
inputting the initial characteristic data into a target attribute value determination model to generate a prediction result corresponding to the prediction target;
the advance period value between each first attribute dimension and the target attribute dimension is obtained according to the following steps:
acquiring historical characteristic data; the history characteristic data comprises attribute values corresponding to each first attribute dimension and the target attribute dimension in a plurality of continuous history acquisition periods;
generating the target attribute under the condition of different acquisition intervals according to the attribute values corresponding to the target attribute dimensions and the attribute values corresponding to each first attribute dimension respectivelyInterval relativity between each attribute dimension and each sex dimension; wherein, under the condition of spacing e acquisition periods, the interval correlation degree R of the target attribute dimension and the b first attribute dimension e b The following relationship is satisfied:
wherein M is a The attribute value corresponding to the a-th acquisition period in the target attribute dimension is obtained; y is Y g The feature mean value corresponding to the target attribute dimension; b (B) a-e The attribute values corresponding to the a-e acquisition periods in the b first attribute dimension are obtained; y is Y b The characteristic mean value corresponding to the b first attribute dimension; z is the number of attribute values contained in the first attribute dimension; e=0, 1, … x, x being the upper interval limit, x<z；
Taking the acquisition cycle interval number corresponding to the correlation extremum of the target attribute dimension and each first attribute dimension as a preceding cycle value between the target attribute dimension and each first attribute dimension; the correlation extremum is the maximum value of the absolute value of the interval correlation between the target attribute dimension and the first attribute dimension.
2. The method of claim 1, wherein before obtaining the attribute value corresponding to the target acquisition period in each first attribute dimension according to the predicted time t corresponding to the predicted attribute value and the advance period value between each first attribute dimension and the target attribute dimension, the method further comprises:
if the advance period value corresponding to the first attribute dimension is greater than zero, determining that the first attribute dimension is one data dimension in the initial feature data.
3. The method of claim 1, wherein before obtaining the attribute value corresponding to the target acquisition period in each first attribute dimension according to the predicted time t corresponding to the predicted attribute value and the advance period value between each first attribute dimension and the target attribute dimension, the method further comprises:
and if the advance period value corresponding to the first attribute dimension is smaller than zero, determining that the first attribute dimension is one data dimension in the verification data.
4. The method of claim 1, wherein obtaining a plurality of first attribute dimensions from a preset feature library according to target attribute dimensions corresponding to a predicted target comprises:
generating a correlation value of the target attribute dimension and each attribute dimension according to the target attribute dimension and a plurality of attribute values corresponding to each attribute dimension in a preset feature library;
acquiring a first attribute dimension from a plurality of attribute dimensions according to the correlation value corresponding to each attribute dimension; the first attribute dimension is an attribute dimension with a correlation value greater than a first correlation threshold.
5. The method of claim 4, generating a correlation value of the target attribute dimension and each attribute dimension according to a plurality of attribute values corresponding to the target attribute dimension and each attribute dimension in a preset feature library, comprising:
and processing a plurality of attribute values corresponding to each attribute dimension in the target attribute dimension and a preset feature library by using a covariance algorithm or a correlation algorithm to generate a correlation value of the target attribute dimension and each attribute dimension.
6. The method of claim 1, wherein obtaining historical feature data comprises:
acquiring an attribute value of each attribute dimension generated in a preset history period;
dividing the preset history period into a plurality of acquisition periods; the acquisition period comprises at least one multi-value dimension, wherein the multi-value dimension is an attribute dimension for acquiring a plurality of attribute values in the corresponding acquisition period;
and generating a target attribute value corresponding to each multi-value dimension according to the average value of the attribute values included in each multi-value dimension.
7. The method of claim 1, wherein the initial model is iteratively calculated using an L-BFGS algorithm during training of the target attribute value determination model.
8. The method of claim 1, wherein the first correlation threshold has a value interval of [0.4,1].
9. A non-transitory computer readable storage medium storing a computer program, wherein the computer program when executed by a processor implements a method of target prediction based on time-series-associated data as claimed in any one of claims 1 to 8.
10. An electronic device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements a method of target prediction based on time-series-related data as claimed in any one of claims 1 to 8 when the computer program is executed by the processor.
CN202311147980.3A 2023-09-07 2023-09-07 Target prediction method, storage medium and device based on time sequence associated data Pending CN116883060A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202311147980.3A CN116883060A (en) 2023-09-07 2023-09-07 Target prediction method, storage medium and device based on time sequence associated data

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311147980.3A CN116883060A (en) 2023-09-07 2023-09-07 Target prediction method, storage medium and device based on time sequence associated data

Publications (1)

Publication Number Publication Date
CN116883060A true CN116883060A (en) 2023-10-13

Family

ID=88272076

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311147980.3A Pending CN116883060A (en) 2023-09-07 2023-09-07 Target prediction method, storage medium and device based on time sequence associated data

Country Status (1)

Country Link
CN (1) CN116883060A (en)

Similar Documents

Publication Publication Date Title
US20190325514A1 (en) Credit risk prediction method and device based on lstm model
Gu et al. Forecasting Nickel futures price based on the empirical wavelet transform and gradient boosting decision trees
US11861464B2 (en) Graph data structure for using inter-feature dependencies in machine-learning
CN112270547A (en) Financial risk assessment method and device based on feature construction and electronic equipment
CN112017061A (en) Financial risk prediction method and device based on Bayesian deep learning and electronic equipment
CN111401940B (en) Feature prediction method, device, electronic equipment and storage medium
CN112668238B (en) Rainfall processing method, rainfall processing device, rainfall processing equipment and storage medium
CN112988840A (en) Time series prediction method, device, equipment and storage medium
CN113034046A (en) Data risk metering method and device, electronic equipment and storage medium
US11568312B2 (en) Augmenting machine learning models to incorporate incomplete datasets
CN115130894A (en) Production planning method and device based on artificial intelligence, computer equipment and medium
CN111260142A (en) Commodity index data prediction method and device, storage medium and electronic equipment
US11544552B1 (en) Method and apparatus for refining an automated coding model
Tang et al. An improved BPNN prediction method based on multi-strategy sparrow search algorithm
CN116883060A (en) Target prediction method, storage medium and device based on time sequence associated data
CN113688202B (en) Emotion polarity analysis method and device, electronic equipment and computer storage medium
CN115936802A (en) Personalized marketing method, device, equipment and storage medium based on user portrait and sequence modeling
CN114266414A (en) Loan amount prediction method, loan amount prediction device, loan amount prediction electronic device, and loan amount prediction medium
US20210182696A1 (en) Prediction of objective variable using models based on relevance of each model
Utkin et al. SurvBeX: An explanation method of the machine learning survival models based on the Beran estimator
CN112200488A (en) Risk identification model training method and device for business object
CN111105274A (en) Data processing method, device, medium and electronic equipment
Li et al. Stock price series forecasting using multi-scale modeling with boruta feature selection and adaptive denoising
CN116402321B (en) Method and device for determining demand of article, electronic equipment and storage medium
Iqbal et al. Forecasting stock market using machine learning approach encoder-decoder ConvLSTM

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