JP6130977B1 - Information processing apparatus, information processing method, information processing system, and program - Google Patents

Abstract

Properly predict future objective variables by multivariate analysis. An information processing apparatus according to an aspect of the present invention includes an acquisition unit that acquires data, a selection unit that selects an objective variable and one or more explanatory variable candidates from the data, and each explanatory variable candidate A construction unit for constructing a series model, wherein the construction unit uses one or more explanatory variables selected from the explanatory variable candidates based on the prediction accuracy of each explanatory variable candidate by the time series model, A multivariate model of the objective variable is constructed.

Description

  The present invention relates to an information processing apparatus, an information processing method, an information processing system, and a program related to data prediction.

  Efforts are being made to collect a large amount of data, analyze changes in the data with time, and make use of the data. For such analysis, multivariate analysis (for example, multiple regression analysis) is widely used.

  For example, a technique has been proposed for constructing a prediction model for predicting time-series data (also referred to as objective variables) that changes over time using another time-series data (also referred to as explanatory variables). (Patent Document 1).

JP 2013-152656 A

  However, in order to predict the future objective variable, it is necessary to consider the future explanatory variable. However, the future explanatory variable cannot be acquired unless it is at that time. For this reason, the conventional multivariate analysis method cannot suitably predict the future target variable.

  The present invention has been made in view of such points, and an object of the present invention is to provide an information processing apparatus, an information processing method, an information processing system, and a program capable of suitably predicting a future target variable by multivariate analysis. One.

An information processing apparatus according to an aspect of the present invention includes an acquisition unit that acquires data, a selection unit that selects an objective variable and one or more explanatory variable candidates from the data, and a first of the data building a time series model for each explanatory variable candidates with the duration of the data, the construction unit for calculating a prediction accuracy of the time-series models for each explanatory variable candidates by using the data of the second period of the data, The construction unit constructs a multivariate model of the objective variable using one or more explanatory variables selected from each explanatory variable candidate based on the prediction accuracy of the time series model of each explanatory variable candidate. It is characterized by that.

  According to the present invention, a future objective variable can be suitably predicted by multivariate analysis.

It is a figure which shows an example of schematic structure of the information processing system which concerns on one Embodiment of this invention. It is a conceptual explanatory drawing of the objective variable prediction method which concerns on one Embodiment of this invention. It is a figure which shows an example of the flowchart of the objective variable prediction method which concerns on one Embodiment of this invention. It is a figure which shows an example of the selection screen of the explanatory variable candidate which concerns on FIG.3 S12. It is a figure which shows an example of the display screen of the graph regarding the explanatory variable and objective variable which concern on FIG.3 S18. It is a figure which shows an example of the function structure of the server which concerns on one Embodiment of this invention. It is a figure which shows an example of the hardware constitutions of the server and device which concern on one Embodiment of this invention.

  The inventor has conceived that the relationship between the objective variable and the explanatory variable is analyzed by a multivariate model, and each explanatory variable is predicted by a time series model. And it discovered that the explanatory variable employ | adopted by a multivariate model was selected appropriately based on the prediction accuracy of the explanatory variable by a time series model.

  As a result, it is possible to restrict the explanatory variable with poor prediction accuracy from being used for prediction of the objective variable, and it is possible to suppress a situation in which the prediction accuracy of the objective variable is greatly deteriorated.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

(Information processing system)
First, an information processing system to which the present invention is applied will be described. FIG. 1 is a diagram illustrating an example of a schematic configuration of an information processing system according to an embodiment of the present invention. An information processing system 1 illustrated in FIG. 1 includes a server 10 and a device 20.

  In the embodiment of the present invention, in the configuration of the information processing system illustrated in FIG. 1, the server 10 performs multivariate analysis on data based on an instruction from the device 20, and the analysis result is transmitted to the device 20. Send.

  The server 10 is an information processing apparatus having a function of analyzing data based on an instruction from the device 20. The server 10 may be composed of a plurality of servers.

  The device 20 is an information processing apparatus that executes an application such as a browser by a user operation and communicates with the server 20. The device 10 may be a mobile terminal (mobile communication terminal) such as a mobile phone, a smartphone, or a tablet terminal, or may be a fixed communication terminal such as a personal computer (PC).

  An example of the functional configuration and hardware configuration of each device such as the device 10 and the server 20 will be described later.

  Note that the system configuration is an example, and the present invention is not limited to this. For example, each device is configured to include one device in FIG. 1, but the number of devices is not limited to this, and a plurality of devices may exist. Further, the information processing system 1 may be configured such that a function of a predetermined device is realized by a plurality of devices.

(Information processing method)
An information processing method (objective variable prediction method) according to an embodiment of the present invention will be described below. Each information processing method may be applied to the information processing system described above.

  FIG. 2 is a conceptual explanatory diagram of an objective variable prediction method according to an embodiment of the present invention. As a premise, the server 10 shown in FIG. 1 (hereinafter simply referred to as “server”) is the actual value of the data (object variable) to be predicted and the time series data that is assumed to affect the data. It is assumed that the actual value of (explanatory variable) is held.

  The server builds a multivariate model based on the actual value of the objective variable and the actual value of one or more explanatory variables (quantifies the relationship of each data). A multivariate model formulates (models) the relationship between objective variables and explanatory variables. For example, a model can be constructed by multiple regression analysis, discriminant analysis, logistic regression analysis, nonlinear analysis, or the like. . Hereinafter, in this specification, a multiple regression model will be described as an example of a multivariate model. However, the present invention can be applied to other models (that is, the term multiple regression model can be read as a multivariate model). May be)

The multiple regression model is represented by the following formula 1, for example.
(Formula 1)
Objective variable = Ax ₁ + Bx ₂ + Cx ₃ + ... + Z
Here, A, B, and C are coefficients obtained by multiple regression analysis, x ₁ , x ₂ , and x ₃ are explanatory variables, and Z is an intercept (constant) obtained by multiple regression analysis.

  Although three explanatory variables are given in FIG. 2, the number of explanatory variables to be used is not limited to this. Further, when only one explanatory variable is selected, a single regression model may be used.

  Further, the server constructs a time series model of each explanatory variable based on the actual value of each explanatory variable. The time series model is a model that can predict future movements from past patterns and trends, and will be described in detail later.

  In the present embodiment, the construction of the time series model is related to the construction of the multiple regression model described above. Specifically, the server narrows down the explanatory variables used for the multiple regression model based on the constructed time series model. For this reason, the server may reconstruct (update) the multiple regression model as appropriate.

  When the time series model and the multiple regression model are determined, the server calculates each explanatory variable prediction value based on each time series model, and inputs these into the multiple regression model to calculate the target variable prediction value.

  Hereinafter, a specific processing flow will be described with reference to FIG. FIG. 3 is a diagram illustrating an example of a flowchart of an objective variable prediction method according to an embodiment of the present invention.

  The server prepares data used for processing (step S11). The server may upload predetermined data (also referred to as source data) from the device 10 or obtain it separately (for example, download from a predetermined server (not shown)). For example, the source data may be expressed in CSV (Comma-Separated Values) format.

  The server creates a variable (processes the source data) based on the acquired source data. Each variable may be used as an objective variable or may be used as an explanatory variable. For example, when the source data is daily data (daily data), the server may convert the data into annual data. A plurality of data (variables) may be merged.

  Next, the server sets an environment (condition) used for each model construction (step S12). For example, the server selects an objective variable to be predicted from the variables created in step S11. Further, the server selects an explanatory variable candidate used for prediction of the selected objective variable. For example, the server may associate (link) each variable in step S11. In this case, in step S12, the explanatory variable used for prediction of the objective variable is determined from the variable associated with the objective variable. You can choose.

  In step S12, the server sets a prediction reference date, a learning period, an evaluation period, and a prediction period. The prediction reference date is a reference date related to model construction and prediction. The learning period is a period including data (learning data, training data) used for model construction. The evaluation period is a period during which accuracy evaluation is performed using the constructed model. That is, data belonging to the evaluation period is used as test data. The prediction period is a period of a target that is actually predicted using the constructed model.

  The learning period can be a period that goes back from the prediction reference date, and the evaluation period can be a period after the prediction reference date. For example, when the prediction reference date is April 2014, the learning period is 60 months, and the evaluation period is 12 months, the learning period corresponds to March 2009 to March 2014, and the evaluation period is 2014. From April to April 2015. Note that these periods may be defined as periods different from the above definitions.

  In step S12, the server selects an algorithm candidate related to construction of a time-series model for predicting explanatory variable candidates. The candidates are, for example, first-order exponential smoothing, second-order exponential smoothing, third-order exponential smoothing (addition), third-order exponential smoothing (multiplication), autoregressive integrated moving average (ARIMA), seasonal autoregressive sum. It is good also as six types of moving averages (SARIMA: Seasonal ARIMA). Note that at least one of these algorithms may not be included. Moreover, it is good also as a structure containing another algorithm.

The server also sets reference parameters related to model construction of explanatory variables and objective variables. The reference parameter may be, for example, the following (1) to (6), but is not limited to these:
(1) Threshold value of information related to prediction accuracy at the time of evaluation of explanatory variable candidates (for example, threshold value of mean absolute percentage error (MAPE)),
(2) Correlation coefficient threshold at the time of explanatory variable candidate evaluation,
(3) An upper limit value of the number of explanatory variables to be selected,
(4) Presence / absence of multiple collinearity test (if present, specify the Variance Inflation Factor (VIF) value used for multiple collinearity test),
(5) Presence / absence of consideration of lag correlation (described later) (if yes, specifies the maximum slide period (eg, number of months) when lag correlation is derived),
(6) Presence / absence of data regularization (if present, further regularization processing such as mixing, primary difference, secondary difference, etc. is designated).

  For these reference parameters, default values may be set. For example, the threshold value of (2) may be 0.2, the VIF value of (4) may be 10, The maximum slide period of (5) may be 12 months. Note that the default values are not limited to these values. Also, in the above (4) to (6), any of presence or absence may be designated by default.

  In addition, when there is a deficiency related to the learning period and / or the evaluation period in the explanatory variable candidate and / or objective variable data (actual value), it is preferable that the server performs the complementing process. For example, time series prediction, least squares, or the like may be used as a complementing method. Further, the server applies the steadying process to at least one of the objective variable and the explanatory variable candidate when the data steadying is effectively set in the above (6).

  The server obtains a correlation (specifically, a correlation coefficient) in the learning period between the objective variable selected in step S12 and the explanatory variable candidate (step S13). Here, when the lag correlation is set to be derived, the server slides (shifts) the explanatory variable candidates in the time direction (for example, the minus direction (past direction)) for a predetermined slide period, and checks the correlation. . Lag correlation is suitable when there is a relationship in which the value of the objective variable varies after a certain period (lag) after the value of the explanatory variable varies. The correlation may be obtained using the evaluation period instead of the learning period and / or the learning period.

  When deriving the lag correlation, the server slides one explanatory variable candidate from 0 month (no slide) to the maximum slide period set in step S12 to calculate the correlation. For example, when the number of selected explanatory variable candidates is 80, the maximum slide period is 12 months, and there is data for each month, the correlation check in step S13 is performed 80 × (12 + 1) = 1040 times. The explanatory variable candidate to which the slide is applied may be used for subsequent modeling as a variable different from the explanatory variable candidate to which the original slide is not applied.

  When the absolute value of the calculated correlation coefficient is equal to or greater than (or larger than) the threshold value set in step S12, the server performs the process of step S14 on the explanatory variable candidates corresponding to the correlation coefficient. In other words, when the absolute value of the correlation coefficient is less than the threshold (below), the corresponding explanatory variable candidate is excluded from being used for the construction of the multiple regression model.

  Next, for each explanatory variable candidate that has passed the correlation check in step S13, the server uses each of the algorithms selected in step S12 to construct a time series model using the learning period data. Prediction and accuracy evaluation are performed (step S14). For example, when six types of algorithms are selected in step S12 and 30 explanatory variable candidates pass the correlation check in step S13, 30 × 6 = 180 prediction calculations are performed. Based on the prediction result and the actual value (test data) of the evaluation period, information on the prediction accuracy (for example, the error rate of prediction) can be calculated.

  For each explanatory variable candidate, the server determines the most accurate algorithm (for example, with a low error rate), and determines (selects) a time series model constructed by the algorithm as a prediction model of the explanatory variable candidate. . For each explanatory variable candidate, a time series model using a different algorithm may be selected. Here, the error rate is assumed to be MAPE, but another index may be used, and in this case, a threshold value of the index may be set in step S12.

  However, for a certain explanatory variable candidate, when the prediction accuracy by each algorithm is all worse than the threshold value of information related to the prediction accuracy set in step S12 (for example, the error rate exceeds the threshold value set in step S12), The explanatory variable candidate is excluded so as not to be used in the construction of the multiple regression model. That is, in steps S13 and S14, the explanatory variables used in the multiple regression model are selected based on the prediction accuracy of the explanatory variable candidates based on the time series model.

  The server constructs a multiple regression model of the objective variable using the explanatory variable candidates that have not been excluded in the previous steps (step S15). A multiple regression model uses a stepwise method, a variable increase method (advance selection method), a variable decrease method (regression elimination method), a round robin method, and the like, using a predetermined standard (for example, Akaike Information Criterion (AIC: Akaike's Information Criterion) can be constructed by determining a combination of explanatory variables that minimizes, for example, the process of building a model using AIC (that is, the process of selecting explanatory variables used for prediction) is an AIC check May be called.

  Note that as a predetermined standard (an index indicating the usefulness of explanatory variables), an F value, a BIC (Bayesian Information Criterion), or the like may be used.

  The server calculates VIF for the multiple regression model constructed in step S15 and checks the correlation between explanatory variables (VIF check) (step S16). Then, when the VIF calculated for two certain explanatory variables is equal to or greater than the VIF value (eg, 10) set in step S12, one explanatory variable is excluded. Thereby, it can suppress that the problem of multiple collinearity arises. Note that if the multiple collinearity inspection is set to no in step S12, step S16 can be skipped.

  The server removes the explanatory variables excluded in step S16 from the explanatory variable combinations that define the multiple regression model determined in step S15, and again determines the combination of explanatory variables that minimizes the predetermined criterion. A multiple regression model is reconstructed (step S17). For example, with respect to a pair of two explanatory variables that did not pass the VIF check in step S16, an AIC check is performed on a model when one is removed from the explanatory variable and a model when the other is removed from the explanatory variable. The model with the smallest AIC may be adopted.

  By performing the processing in the order of steps S13 to S17, the number of explanatory variable candidates can be sequentially reduced, so that the prediction accuracy can be ensured while suitably reducing the amount of calculation required for model construction.

  The server inputs the predicted value calculated by the time series model of each explanatory variable constructed in step S14 to the multiple regression model completed in step S17, and calculates the predicted value of the objective variable for the prediction period (step S18). ).

  The server obtains information about the multiple regression model completed in step S17, the time series model of each explanatory variable (candidate) created in step S14, the explanatory variable calculated in step S18 and / or the predicted value of the objective variable, from the device 10. May be sent to. For example, the server may transmit a CSV file related to the explanatory variable (candidate) and / or the predicted value of the objective variable to the device 10.

  Further, the server may generate a graph that displays the actual value and the predicted value regarding the explanatory variable (may be an explanatory variable candidate) and / or the objective variable, and may transmit the graph to the device 10 to display the graph. In addition, when the above prediction process is performed with the data regularization enabled, the regularized data may be displayed as the original data or may be displayed as the regularized data. .

  The server may construct a plurality of multiple regression models in steps S15 to S17. In this case, the server may obtain a prediction result when each model is used in step S18.

  The server may communicate with the device 10 so as to display the operation screen related to the flow of FIG. 3, and may control each process in response to a user input or the like in the device 10. FIG. 4 is a diagram showing an example of the explanatory variable candidate selection screen according to step S12 of FIG. FIG. 4 shows a display screen 400 displayed on the device 10.

  The display screen 400 may include a screen switching tab portion 401. When a predetermined tab is selected by a user operation, the information display / operation area 411 may be switched to the corresponding content. In FIG. 4, an explanatory variable selection tab is designated.

  In the area 411, explanatory variable candidates (parameters) used for prediction of objective variables (here, data called commodity future A) are listed. By the user's operation, all of these candidates may be selected as candidates for model construction, or some candidates may not be selected so as to be excluded from the beginning.

  FIG. 5 is a diagram illustrating an example of a graph display screen regarding the explanatory variable and the objective variable according to step S18 of FIG. In the tab part 401, a tab of the prediction result is designated.

  The area 421 is used to reduce and display the actual values and predicted values of the explanatory variable (E) and the objective variable (O). One or more variables can be enlarged and displayed in the region 422 by a user operation. In FIG. 5, the check box of the product future A that is the objective variable is checked, and the predicted value and the actual value of the product future A are shown in the area 422.

  According to the embodiment of the information processing method described above, it is possible to limit the explanatory variable with poor prediction accuracy so as not to be used for prediction of the objective variable, and to suppress a situation in which the prediction accuracy of the objective variable is greatly deteriorated. it can.

  Note that the server may construct a prediction model by a method other than the objective variable prediction method (time series model + multiple regression model) according to the present embodiment. For example, a prediction model construction based on a state space model, a neural network (for example, a recurrent neural network) model, or the like may be set.

(Device configuration)
FIG. 6 is a diagram illustrating an example of a functional configuration of a server according to an embodiment of the present invention. The server 10 includes a control unit 11, a storage unit 12, a communication unit 13, an input unit 14, and an output unit 15. In this example, the functional blocks of the characteristic part in the present embodiment are mainly shown, and the server 10 may also have other functional blocks necessary for other processing. Moreover, it is good also as a structure which does not include one part functional block.

  The control unit 11 controls the server 10. The control part 11 can be comprised by the controller, the control circuit, or control apparatus demonstrated based on the common recognition in the technical field which concerns on this invention. The control unit 11 can configure an acquisition unit, a selection unit, a construction unit, and the like according to an embodiment of the present invention.

  For example, the control unit 11 may perform control that functions as an acquisition unit that acquires data via the storage unit 12, the communication unit 13, the input unit 14, and the like. The control unit 11 may perform control that functions as a selection unit that selects an objective variable and one or more explanatory variable candidates from the acquired data. The control unit 11 may perform control that functions as a construction unit that constructs a time-series model of each explanatory variable candidate.

  The control unit (construction unit) 11 constructs a multivariate model of the objective variable using one or more explanatory variables selected from the explanatory variable candidates based on the prediction accuracy of each explanatory variable candidate based on the time series model. You may control.

  The control unit (construction unit) 11 calculates the prediction accuracy of the time-series model of the predetermined explanatory variable candidate using the test data, and when the prediction accuracy is worse than the predetermined threshold, the predetermined explanatory variable candidate is You may control not to use for the explanatory variable of a multivariate model.

  The control unit (construction unit) 11 performs control so as not to use the corresponding predetermined explanatory variable candidate as the explanatory variable of the multivariate model when predetermined information (for example, MAPE) regarding the prediction accuracy exceeds a predetermined threshold. May be.

  In addition, the control unit (selection unit) 11 may perform control so that the explanatory variable candidate is selected from data having an absolute value of a correlation coefficient equal to or greater than a predetermined value with data corresponding to the target variable. .

  Further, the control unit (selection unit) 11 may perform control so that the data corresponding to the explanatory variable candidate is slid in the time direction as compared with the data corresponding to the target variable and the correlation is obtained.

  The control unit (construction unit) 11 may perform control so as to construct a multivariate model based on a predetermined criterion (for example, AIC).

  The control unit (construction unit) 11 once constructs a multivariate model based on a predetermined standard (for example, AIC), and then excludes at least one of combinations of explanatory variables in which VIF is equal to or greater than a predetermined VIF value, The multivariate model may be controlled to be reconstructed.

  The control unit 11 is preferably capable of performing transaction data processing and / or analysis data processing (online analysis) in real time, and for example, preferably performs control so as to perform in-memory data processing.

  The storage unit 12 stores (holds) information used by the server 10. For example, the storage unit 12 sequentially stores data that is the basis of the objective variable / explanatory variable. The storage unit 12 can be configured by, for example, a memory, a storage, a storage device, or the like described based on common recognition in the technical field according to the present invention.

  The communication unit 13 communicates various information with the device 20. The communication part 13 can be comprised by the transmitter / receiver, the transmission / reception circuit, or transmission / reception apparatus demonstrated based on the common recognition in the technical field which concerns on this invention. The communication unit 13 may include a transmission unit and a reception unit.

  The input unit 14 receives an input by an operation from the user. The input unit 14 may be connected to a predetermined device or storage medium and accept data input. The input unit 14 may output the input result to the control unit 11, for example.

  The input unit 14 can be configured by an input device such as a keyboard, a mouse, and a button, an input / output terminal, an input / output circuit, and the like described based on common recognition in the technical field according to the present invention. Further, the input unit 14 may have a configuration (for example, a touch panel) integrated with the display unit.

  The output unit 15 outputs various information so that the user can recognize it. For example, the output unit 15 may include a display unit that displays an image, an audio output unit that outputs audio, and the like. The display unit can be configured by, for example, a display device such as a display or a monitor described based on common recognition in the technical field according to the present invention. The audio output unit can be configured by an output device such as a speaker described based on common recognition in the technical field according to the present invention.

  The output unit 15 includes, for example, an arithmetic unit, an arithmetic circuit, an arithmetic device, a player, an image / video / audio processing circuit, an image / video / audio processing device, which are described based on common recognition in the technical field according to the present invention, An amplifier or the like can be included.

  The device 20 may have the same configuration as that in FIG. The server 10 may be realized by a cloud platform, and provides a service called SaaS (Software as a Service), PaaS (Platform as a Service), IaaS (Infrastructure as a Service), or the like. May be.

(Hardware configuration)
In addition, the block diagram used for description of the said embodiment has shown the block of the functional unit. These functional blocks (components) are realized by any combination of hardware and / or software. Further, the means for realizing each functional block is not particularly limited. That is, each functional block may be realized by one physically coupled device, or may be realized by two or more physically separated devices connected by wire or wirelessly and by a plurality of these devices. Good.

  For example, the server 10 or the like according to an embodiment of the present invention may function as a computer that performs processing of the information communication method of the present invention. FIG. 7 is a diagram illustrating an example of a hardware configuration of a server and a device according to an embodiment of the present invention. The above-described server 10, device 20, and the like may be physically configured as a computer device including a processor 1001, a memory 1002, a storage 1003, a communication device 1004, an input device 1005, an output device 1006, a bus 1007, and the like.

  In the following description, the term “apparatus” can be read as a circuit, a device, a unit, or the like. The hardware configuration of the server 10 and the device 20 may be configured to include one or a plurality of each device illustrated in the figure, or may be configured not to include some devices.

  For example, although only one processor 1001 is shown, there may be a plurality of processors. Further, the processing may be executed by one processor, or the processing may be executed by one or more processors simultaneously, sequentially, or in another manner.

  Each function in the server 10, the device 20, and the like is performed by causing the processor 1001 to perform calculations by reading predetermined software (programs) on hardware such as the processor 1001 and the memory 1002, and the communication by the communication device 1004 or the memory 1002. This is realized by controlling reading and / or writing of data in the storage 1003.

  For example, the processor 1001 controls the entire computer by operating an operating system. The processor 1001 may be configured by a central processing unit (CPU) including an interface with peripheral devices, a control device, an arithmetic device, a register, and the like. Each unit such as the control unit 11 described above may be realized by the processor 1001. The processor 1001 may be implemented by one or more chips.

  Further, the processor 1001 reads a program (program code), software module, and data from the storage 1003 and / or the communication device 1004 to the memory 1002, and executes various processes according to these. As the program, a program that causes a computer to execute at least a part of the operations described in the above embodiments is used. For example, the control unit 11 may be realized by a control program stored in the memory 1002 and operated by the processor 1001, and may be realized similarly for other functional blocks.

  The memory 1002 is a computer-readable recording medium. For example, the memory 1002 includes at least a ROM (Read Only Memory), an EPROM (Erasable Programmable ROM), an EEPROM (Electrically EPROM), a RAM (Random Access Memory), and other suitable storage media. It may be configured by one. The memory 1002 may be called a register, a cache, a main memory (main storage device), or the like. The memory 1002 can store a program (program code), a software module, and the like that can be executed to perform the information processing method according to an embodiment of the present invention.

  The storage 1003 is a computer-readable recording medium such as a flexible disk, a floppy (registered trademark) disk, a magneto-optical disk (for example, a compact disk (CD-ROM (Compact Disc ROM), etc.)), a digital versatile disk, Blu-ray (registered trademark) disk, removable disk, hard disk drive, smart card, flash memory device (eg, card, stick, key drive), magnetic stripe, database, server, or other suitable storage medium It may be constituted by. The storage 1003 may be referred to as an auxiliary storage device. Note that the above-described storage unit 12 may be realized by the memory 1002 and / or the storage 1003.

  The communication device 1004 is hardware (transmission / reception device) for performing communication between computers via a wired and / or wireless network, and is also referred to as a network device, a network controller, a network card, a communication module, or the like. Note that the communication unit 13 described above may be realized by the communication device 1004.

  The input device 1005 is an input device (for example, a keyboard, a mouse, etc.) that accepts external input. The output device 1006 is an output device (for example, a display, a speaker, etc.) that performs output to the outside. The input device 1005 and the output device 1006 may have an integrated configuration (for example, a touch panel). The input unit 14 and the output unit 15 described above may be realized by the input device 1005 and the output device 1006, respectively.

  Each device such as the processor 1001 and the memory 1002 is connected by a bus 1007 for communicating information. The bus 1007 may be configured with a single bus or may be configured with different buses between apparatuses.

  The television 10, the device 20, the matching server 40, and the like include a microprocessor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a programmable logic device (PLD), and a field programmable gate array (FPGA). Or a part of each functional block may be realized by the hardware. For example, the processor 1001 may be implemented by at least one of these hardware.

(Modification)
Note that the terms described in this specification and / or terms necessary for understanding this specification may be replaced with terms having the same or similar meaning.

  Information, parameters, and the like described in the present specification may be represented by absolute values, may be represented by relative values from predetermined values, or may be represented by other corresponding information. In addition, names used for parameters and the like in this specification are not limited in any respect.

  Information, signals, etc. described herein may be represented using any of a variety of different technologies. For example, data, commands, commands, information, signals, bits, symbols, chips, etc. that may be referred to throughout the above description are voltages, currents, electromagnetic waves, magnetic fields or magnetic particles, light fields or photons, or any of these May be represented by a combination of

  Information, signals, and the like may be input / output via a plurality of network nodes. Input / output information, signals, and the like may be stored in a specific location (for example, a memory) or may be managed in a table. Input / output information, signals, and the like can be overwritten, updated, or added. The output information, signals, etc. may be deleted. Input information, signals, and the like may be transmitted to other devices.

  In addition, notification of predetermined information (for example, notification of being “X”) is not limited to explicitly performed, but is performed implicitly (for example, by not performing notification of the predetermined information). May be.

  Software, whether it is called software, firmware, middleware, microcode, hardware description language, or other names, instructions, instruction sets, codes, code segments, program codes, programs, subprograms, software modules , Applications, software applications, software packages, routines, subroutines, objects, executable files, execution threads, procedures, functions, etc. should be interpreted broadly.

  Further, software, instructions, information, and the like may be transmitted / received via a transmission medium. For example, software may use websites, servers, or other devices using wired technology (coaxial cable, fiber optic cable, twisted pair and digital subscriber line (DSL), etc.) and / or wireless technology (infrared, microwave, etc.) When transmitted from a remote source, these wired and / or wireless technologies are included within the definition of transmission media.

  As used herein, the terms “system” and “network” are used interchangeably.

  Each aspect / embodiment described in this specification may be used independently, may be used in combination, or may be switched according to execution. In addition, the order of the processing procedures, sequences, flowcharts, and the like of each aspect / embodiment described in this specification may be changed as long as there is no contradiction. For example, the methods described herein present the elements of the various steps in an exemplary order and are not limited to the specific order presented.

  As used herein, the phrase “based on” does not mean “based only on,” unless expressly specified otherwise. In other words, the phrase “based on” means both “based only on” and “based at least on.”

  Although the present invention has been described in detail above, it will be apparent to those skilled in the art that the present invention is not limited to the embodiments described herein. The present invention can be implemented as modified and changed modes without departing from the spirit and scope of the present invention defined by the description of the scope of claims. Therefore, the description of the present specification is for illustrative purposes and does not have any limiting meaning to the present invention.

Claims (12)

An acquisition unit for acquiring data;
A selection unit for selecting an objective variable and one or more explanatory variable candidates from the data;
Among the data, the time series model of each explanatory variable candidate is constructed using the data of the first period, and the prediction accuracy of the time series model of each explanatory variable candidate is set using the data of the second period of the data. A construction unit for calculating ,
The construction unit constructs a multivariate model of the objective variable using one or more explanatory variables selected from each explanatory variable candidate based on the prediction accuracy of the time series model of each explanatory variable candidate. Information processing apparatus. A control unit that inputs, as the one or more explanatory variables, a predicted value calculated by a time series model of each explanatory variable to the multivariate model and calculates a predicted value of the objective variable. Item 4. The information processing apparatus according to Item 1. The building unit, if the prediction accuracy of the time series model in a predetermined explanatory variable candidates is worse than a predetermined threshold, the predetermined explanatory variable candidates, is characterized by not using the explanatory variable of the multivariate model The information processing apparatus according to claim 1. The said selection part selects each explanatory variable candidate from the data which have the absolute value of a correlation coefficient more than predetermined value between the data corresponding to the said objective variable, The Claim 1 characterized by the above-mentioned. 4. The information processing apparatus according to any one of 3. The information processing apparatus according to claim 4, wherein the selection unit obtains a correlation by sliding data corresponding to each explanatory variable candidate in a time direction as compared with data corresponding to the objective variable.   The time series model includes first-order exponential smoothing, second-order exponential smoothing, third-order exponential smoothing (addition), third-order exponential smoothing (multiplication), autoregressive integrated moving average (ARIMA) or seasonal autoregressive sum. 6. The information processing apparatus according to claim 1, wherein the information processing apparatus is at least one of a moving average (SARIMA: Seasonal ARIMA).   The information processing apparatus according to claim 1, wherein the multivariate model is a multiple regression model.   The information processing apparatus according to any one of claims 1 to 7, wherein the construction unit constructs the multivariate model based on an Akaike's Information Criterion (AIC).   The construction unit temporarily constructs the multivariate model based on the AIC, and then excludes at least one of combinations of explanatory variables in which a variance expansion factor (VIF) is a predetermined VIF value or more and excludes the multivariate model. The information processing apparatus according to claim 8, wherein a variable model is reconstructed. Acquiring data;
Selecting an objective variable and one or more explanatory variable candidates from the data;
Among the data, the time series model of each explanatory variable candidate is constructed using the data of the first period, and the prediction accuracy of the time series model of each explanatory variable candidate is set using the data of the second period of the data. A calculating step,
An information processing method characterized in that a multivariate model of the objective variable is constructed using one or more explanatory variables selected from each explanatory variable candidate based on the prediction accuracy of the time series model of each explanatory variable candidate. An information processing system including an information processing device,
The information processing apparatus includes an acquisition unit that acquires data;
A selection unit for selecting an objective variable and one or more explanatory variable candidates from the data;
Among the data, the time series model of each explanatory variable candidate is constructed using the data of the first period, and the prediction accuracy of the time series model of each explanatory variable candidate is set using the data of the second period of the data. A construction unit for calculating ,
The construction unit constructs a multivariate model of the objective variable using one or more explanatory variables selected from each explanatory variable candidate based on the prediction accuracy of the time series model of each explanatory variable candidate. Information processing system. On the computer,
The steps to get the data,
A procedure for selecting an objective variable and one or more explanatory variable candidates from the data;
Among the data, the time series model of each explanatory variable candidate is constructed using the data of the first period, and the prediction accuracy of the time series model of each explanatory variable candidate is set using the data of the second period of the data. The procedure to calculate ,
A program for executing a procedure for constructing a multivariate model of the objective variable using one or more explanatory variables selected from each explanatory variable candidate based on the prediction accuracy of the time series model of each explanatory variable candidate .

Images