Cause-and-effect analysis support device, method for supporting analysis of cause and effect, program, and recording medium

The causal analysis support device enhances model readability and fitness by acquiring and reconstructing causal models based on variable importance, addressing the complexity of multivariate causal analysis.

JP2024171826APending Publication Date: 2024-12-12NEC SOLUTION INNOVATORS LTD
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
JP2023089069
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Filing Date
2023-05-30
Publication Date
2024-12-12

AI Technical Summary

Technical Problem

When performing causal analysis with a large number of variables, the complexity of causes and effects makes it difficult to interpret the analysis results, and setting a threshold for causal coefficients can lead to a decrease in the fitness of the analytical model.

Method used

A causal analysis support device that includes a causal model acquisition unit, importance calculation unit, and reconstruction unit to acquire, calculate the importance of variables, and reconstruct the causal model based on importance levels, generating reconstructed models with improved readability and fitness.

Benefits of technology

The device suppresses the decrease in fitness and improves the readability of the causal model by calculating variable importance and reconstructing models, suitable for fields requiring causal relationship analysis.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 2024171826000001_ABST
    Figure 2024171826000001_ABST
Patent Text Reader

Abstract

To provide a cause-and-effect analysis support device in which the degree of conformance is prevented from dropping and the readability of a cause-and-effect model can be improved.SOLUTION: The cause-and-effect analysis support device includes a cause-and-effect model acquisition unit, an importance degree calculation unit, a re-configuration unit, and a result output unit. The cause-and-effect model acquisition unit acquires a cause-and-effect model including a cause-and-effect relation with at least one variable. The importance degree calculation unit calculates the importance degree of the variable of the cause-and-effect model. The re-configuration unit re-configures the cause-and-effect model on the basis of the importance degree and generates at least one re-configuration model. The result output unit outputs at least one of the re-configuration model.SELECTED DRAWING: Figure 1
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present disclosure relates to a causal analysis support device, a causal analysis support method, a program, and a recording medium. [Background technology]

[0002] In recent years, attempts have been made to estimate the causal relationships of events consisting of multiple variables in order to estimate the causes of various events. For example, Patent Document 1 describes a technology for obtaining information on the causal relationships between mechanisms that accurately reflects the actual situation of a production line (Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2020-149303 A Summary of the Invention [Problem to be solved by the invention]

[0004] However, when performing an analysis using a large number of variables using a system such as that described in Patent Document 1, the number of causal factors between the variables becomes enormous, making it difficult to interpret the analysis results. In this case, for example, if a threshold is set for the causal coefficient to suppress the number of causal factors, there is a problem that the goodness of fit (Comparative Fit Index: CFI) of the analysis model decreases.

[0005] Therefore, an object of the present disclosure is to provide a causal analysis support device that can suppress a decrease in the degree of conformance and improve the readability of a causal model. [Means for solving the problem]

[0006] In order to achieve the above object, the causal analysis support device of the present disclosure comprises: The present invention includes a causal model acquisition unit, an importance calculation unit, a reconstruction unit, and a result output unit, the causal model acquisition unit acquires a causal model including a causal relationship of at least one variable; The importance calculation unit calculates the importance of the variables included in the causal model; The reconstructing unit reconstructs the causal model based on the importance to generate at least one reconstructed model; The result output unit outputs at least one of the reconstructed models.

[0007] The causal analysis support method of the present disclosure includes: The method includes a causal model acquisition step, a importance calculation step, a reconstruction step, and a result output step, The causal model acquisition step acquires a causal model including a causal relationship of at least one variable; The importance calculation step calculates the importance of the variables included in the causal model, The reconstructing step reconstructs the causal model based on the importance to generate at least one reconstructed model; The result output step outputs at least one of the reconstructed models.

[0008] The program of the present disclosure is It includes a procedure for obtaining a causal model, a procedure for calculating importance, a procedure for reconstructing the model, and a procedure for outputting the results. The causal model acquisition step acquires a causal model including a causal relationship of at least one variable; The importance calculation step calculates the importance of the variables included in the causal model; The reconstruction step reconstructs the causal model based on the importance to generate at least one reconstructed model; The result output step outputs at least one of the reconstructed models; The program is for causing a computer to execute each of the above procedures.

[0009] The recording medium of the present disclosure includes: It includes a procedure for obtaining a causal model, a procedure for calculating importance, a procedure for reconstructing the model, and a procedure for outputting the results. The causal model acquisition step acquires a causal model including a causal relationship of at least one variable; The importance calculation step calculates the importance of the variables included in the causal model; The reconstruction step reconstructs the causal model based on the importance to generate at least one reconstructed model; The result output step outputs at least one of the reconstructed models; A computer-readable recording medium having recorded thereon a program for causing a computer to execute each of the above procedures. Effect of the Invention

[0010] According to the present disclosure, it is possible to suppress a decrease in the goodness of fit of a causal model and to improve the readability of the causal model. [Brief description of the drawings]

[0011] [Figure 1] FIG. 1 is a block diagram showing a configuration of an example of a causal analysis support device according to the present disclosure. [Diagram 2] FIG. 2 is a block diagram illustrating an example of a hardware configuration of the causal analysis support device according to the present disclosure. [Diagram 3] FIG. 3 is a flowchart showing an example of processing in the causal analysis support device of the present disclosure. [Figure 4] FIG. 4 is a schematic diagram showing an example of a causal model. [Diagram 5] FIG. 5 is a flowchart illustrating an example of processing by the reconstruction unit of the causal analysis support device of the present disclosure. [Figure 6] FIG. 6 is a diagram showing the evaluation causal model and its degree of conformance in the first embodiment. [Figure 7] FIG. 7 is a diagram showing the reconstructed model and its degree of conformance in the first embodiment. [Figure 8] FIG. 8 is a diagram showing a causal model of a reference example in the first embodiment and its degree of conformance. [Figure 9] FIG. 9 is a block diagram illustrating a configuration of an example of a causal analysis support device according to the present disclosure. [Figure 10] FIG. 10 is a flowchart showing an example of processing in the causal analysis support device of the present disclosure. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0012] Next, an embodiment of the present disclosure will be described with reference to the drawings. The present disclosure is not limited to the following embodiment. In each of the following drawings, the same parts are given the same reference numerals. In addition, the description of each embodiment can be mutually incorporated unless otherwise specified, and the configurations of each embodiment can be combined unless otherwise specified.

[0013] [Embodiment 1] The causal analysis support device of this embodiment will be described with reference to Fig. 1. Fig. 1 is a block diagram showing the configuration of an example of a causal analysis support device 10 of this embodiment. As shown in Fig. 1, the causal analysis support device 10 (hereinafter also referred to as "this device 10") includes a causal model acquisition unit 11, an importance calculation unit 12, a reconstruction unit 13, and a result output unit 14. Although not shown, this device 10 may also include, for example, a storage unit.

[0014] The device 10 may be, for example, a single device including each of the above-mentioned units, or a device to which each of the above-mentioned units can be connected via a communication line network. The device 10 may be connected to an external device described later via a communication line network. The communication line network is not particularly limited, and a known network may be used, for example, wired or wireless. Examples of the communication line network include an Internet line, a World Wide Web (WWW), a telephone line, a Local Area Network (LAN), a Storage Area Network (SAN), a Delay Tolerant Networking (DTN), a Low Power Wide Area (LPWA), and a Local 5G (L5G). Examples of wireless communication include Wi-Fi (registered trademark), Bluetooth (registered trademark), a Local 5G, and a LPWA. Examples of the wireless communication may include a form in which each device directly communicates with each other (Ad Hoc communication), an infrastructure (infrastructure communication), and an indirect communication via an access point. The device 10 may be incorporated into a server as a system, for example. The device 10 may be, for example, a personal computer (PC, for example, desktop type or notebook type) on which the program of the present disclosure is installed, a smartphone, a tablet terminal, etc. Furthermore, the device 10 may be in the form of cloud computing or edge computing, for example, in which at least one of the units is on a server and the other units are on a terminal.

[0015] 2 illustrates a block diagram of a hardware configuration of the present device 10. The present device 10 includes, for example, a central processing unit (CPU, GPU, etc.) 101, a memory 102, a bus 103, a storage device 104, an input device 105, an output device 106, a communication device 107, etc. Each part of the present device 10 is connected to each other via the bus 103 by each interface (I / F).

[0016] The central processing unit 101 cooperates with other components through a controller (such as a system controller or an I / O controller) and controls the entire device 10. In the device 10, the central processing unit 101 executes, for example, the program disclosed herein and other programs, and also reads and writes various information. Specifically, for example, the central processing unit 101 functions as a causal model acquisition unit 11, an importance calculation unit 12, a reconstruction unit 13, and a result output unit 14. The device 10 may include, as a computing device, other computing devices such as a CPU, a GPU (Graphics Processing Unit), an APU (Accelerated Processing Unit), or a combination of these.

[0017] The bus 103 can also be connected to, for example, an external device. Examples of the external device include a device capable of generating a causal model described later, an external storage device (external database, etc.), a printer, an external input device, an external display device, an audio output device such as a speaker, an external imaging device such as a camera, and various sensors such as an acceleration sensor, a geomagnetic sensor, and a direction sensor. The present device 10 can be connected to an external network (the above-mentioned communication line network) by, for example, a communication device 107 connected to the bus 103, and can also be connected to other devices such as a user's terminal via the external network.

[0018] An example of the memory 102 is a main memory (primary storage device). When the central processing unit 101 performs processing, the memory 102 reads various operation programs, such as the program of the present disclosure, stored in the storage device 104 described below, and the central processing unit 101 receives data from the memory 102 and executes the program. The main memory is, for example, a RAM (random access memory). The memory 102 may also be, for example, a ROM (read only memory).

[0019] The storage device 104 is also called an auxiliary storage device, for example, in contrast to the main memory (primary storage device). As described above, the storage device 104 stores an operating program including the program of the present disclosure. The storage device 104 may be, for example, a combination of a recording medium and a drive that reads and writes from the recording medium. The recording medium is not particularly limited, and may be, for example, an internal or external type, such as a HD (hard disk), CD-ROM, CD-R, CD-RW, MO, DVD, flash memory, and memory card. The storage device 104 may be, for example, a hard disk drive (HDD) in which a recording medium and a drive are integrated, and a solid state drive (SSD). When the present device 10 includes the storage unit, for example, the storage device 104 functions as the storage unit.

[0020] In the present device 10, the memory 102 and the storage device 104 can also store various information such as log information, information acquired from an external database (not shown) or an external device, information generated by the present device 10, and information used when the present device 10 executes processing. In this case, the memory 102 and the storage device 104 may store, for example, the above-mentioned information on the user of the present device. Note that at least a portion of the information may be stored, for example, in an external server other than the memory 102 and the storage device 104, or may be distributed and stored in multiple terminals using block chain technology or the like.

[0021] The device 10 further includes, for example, an input device 105 and an output device 106. The input device 105 may be, for example, a pointing device such as a touch panel, a track pad, or a mouse; a keyboard; an imaging means such as a camera or a scanner; a card reader such as an IC card reader or a magnetic card reader; or an audio input means such as a microphone. The output device 106 may be, for example, a display device such as an LED display or a liquid crystal display; an audio output device such as a speaker; a printer; or the like. In this embodiment, the input device 105 and the output device 106 are configured separately, but the input device 105 and the output device 106 may be configured as an integrated device, such as a touch panel display.

[0022] Next, an example of the causal analysis support method of this embodiment will be described with reference to the flowchart of Fig. 3. The causal analysis support method of this embodiment can be implemented as follows, for example, using the causal analysis support device 10 shown in Fig. 1 or Fig. 2. Note that the causal analysis support method of this embodiment is not limited to use of the causal analysis support device 10 of Fig. 1 or Fig. 2. Fig. 3 is a flowchart showing an example of processing by the causal analysis support device 10.

[0023] First, the causal model acquisition unit 11 acquires a causal model including a causal relationship of at least one variable (S1, causal model acquisition step). The causal model is, for example, a model showing the result of searching for the presence or absence of a causal relationship and its direction in at least one variable (also called a factor) by a known causal search method. The causal model includes an arrow (causality) indicating a causal relationship between variables, and a causal coefficient indicating the influence of the causality. The causal search method is not particularly limited, and examples thereof include LiNGAM (Linear Non-Gaussian Acyclic Model), SAM (Structural Agnostic Modelling), Bayesian network, LiNG (Linear Non-Gaussian SEM), and the like. For example, as shown in FIG. 4, the causal model can be expressed as a directed graph in which causal relationships are indicated by arrows for at least one variable (for example, variables A, B, C, and D). In the causal model shown in FIG. 4, for example, variable B has a causal relationship with variables C and D, and variable A has a causal relationship with variable D. In the causal model shown in FIG. 4, the causal coefficient between each variable can be shown as a number near the arrow indicating the causal relationship. Specifically, in the causal model shown in FIG. 4, the causal coefficient from variable A to variable D is -0.2, the causal coefficient from variable B to variable C is 0.6, the causal coefficient from variable B to variable D is -2.6, and the causal coefficient from variable C to variable D is -2.3. The causal model acquisition unit 11 may acquire, for example, one causal model, or may acquire two or more. The causal model acquisition unit 11 may acquire, for example, a causal model analyzed by an external device capable of causal search, or may acquire the causal model from an external database or server in which the causal model is recorded. The causal model acquisition unit 11 may acquire, for example, a causal model generated by the present device 10. In this case, the present device 10 may acquire, for example, data including at least one variable, analyze the acquired data using the causal search method, estimate the causal relationship of the variables included in the input data, and generate the causal model. At this time, for example, if the input data includes one variable, the present device 10 may end the process.At this time, for example, the result output unit 14 described later may also output a result indicating that the number of variables is one.

[0024] The causal model may be, for example, a model in which a target variable is specified, or may be a model in which a target variable is not specified. The target variable means, for example, a variable that is the start or end point of causation. When the target variable is a variable that is the start point of causation, for example, in the causal exploration method, a causal model in which the target variable is the start point can be generated by setting parameters so that causation occurs from the target variable to other variables. Also, when the target variable is a variable that is the end point of causation, for example, in the causal exploration method, a causal model in which the target variable is the end point can be generated by setting parameters so that causation converges from other variables to the target variable.

[0025] Next, the importance calculation unit 12 calculates the importance of the variable included in the causal model (S2, importance calculation step). The importance is, for example, an index showing the influence of a variable included in the causal model on other variables. The importance calculation unit 12 can calculate the importance, for example, based on the causal coefficient of the variable. The causal coefficient is not necessarily a positive value, and may include a negative value. When the causal coefficient is a negative value, it is estimated that the variable has a large negative influence on other variables. For this reason, it is preferable that the importance calculation unit 12 replaces the causal coefficient with an absolute value, for example, and calculates the importance based on the absolute value of the causal coefficient. The importance calculation unit 12 can calculate the centrality index of the variable as the importance. Specifically, the importance calculation unit 12 can perform a network centrality analysis with each variable included in the causal model as a node and the causal coefficient (or its absolute value) as an edge, and calculate the centrality index of each variable as the importance. The means of network centrality analysis is not particularly limited, and for example, the PageRank algorithm, closeness centrality, degree centrality, eigenvector centrality, Bonacic's power centrality, betweenness centrality, information centrality, etc. can be used.

[0026] A specific example of the importance calculation step will be described with reference to FIG. 5. In the following description, as shown in FIG. 5(A), in the causal model shown in FIG. 4, the causal coefficient from variable A to variable D is −0.2, the causal coefficient from variable B to variable C is 0.6, the causal coefficient from variable B to variable D is −2.6, and the causal coefficient from variable C to variable D is −2.3. The importance calculation unit 12 replaces each causal coefficient with an absolute value, for example, as shown in FIG. 5(B). Then, the importance calculation unit 12 performs, for example, a network centrality analysis, and calculates the centrality index of each variable as the importance. FIG. 5(C) is a table showing the importance of variables A to D. As shown in FIG. 5(C), the importance of variable A in the causal model shown in FIG. 4 is 0.83, the importance of variable B is 3.45, the importance of variable C is 2.89, and the importance of variable D is 7.75. For this reason, in the causal model shown in Figure 4, it can be seen that the influence on other variables is greatest in the following order: variable D, variable B, variable C, and variable A.

[0027] Next, the reconstruction unit 13 reconstructs the causal model based on the importance to generate at least one reconstructed model (S3, reconstruction step). Specifically, the reconstruction unit 13 generates candidate variables by deleting at least one variable from the variables included in the causal model based on the importance. The reconstruction unit 13 may, for example, generate the candidate variables by deleting a variable with the lowest importance, or may delete a variable with the importance less than a threshold. The threshold is not particularly limited and can be set appropriately according to the purpose of analysis, etc. The reconstruction unit 13 generates at least one reconstructed model based on the candidate variables, for example. The reconstruction unit 13 can generate the reconstructed model by analyzing the causal relationship between the candidate variables by a known causal search method for the candidate variables, for example. The causal search method is not particularly limited, and examples thereof include LiNGAM (Linear Non-Gaussian Acyclic Model), SAM (Structural Agnostic Modelling), Bayesian network, LiNG (Linear Non-Gaussian SEM), etc. The reconstruction unit 13 may also calculate, for example, the goodness of fit of the generated reconstructed model. Examples of the goodness of fit include CFI (Comparative Fit Index), RMSEA (Root Mean Square Error of Approximation), GFI (Goodness of Fit Index), AGFI (Adjusted Goodness of Fit Index), AIC (Akaike's Information Criterion), CAIC (Consistent Akaike's Information Criterion), and BIC (Bayesian Information Criterion). The device 10 may store the generated reconstructed model in the storage unit. In this case, it is preferable to store the goodness of fit in association with the reconstructed model.

[0028] The reconstruction unit 13 may generate at least one reconstructed model with the target variable as the end point, may generate at least one reconstructed model with the target variable as the start point, or may generate at least one reconstructed model without specifying the target variable. The target variable means, for example, a variable that is the start point or end point of causality. When the target variable is a variable that is the start point of causality, the reconstruction unit 13 can generate a reconstructed model with the target variable as the start point by, for example, setting parameters in the causal search method so that causality occurs from the target variable to other variables. When the target variable is a variable that is the end point of causality, the reconstruction unit 13 can generate a reconstructed model with the target variable as the end point by, for example, setting parameters in the causal search method so that causality converges from other variables to the target variable. When generating a reconstructed model without specifying the target variable, the reconstruction unit 13 can generate a causal model that is the center of the at least one variable for the variables included in the causal model acquired in S1.

[0029] Furthermore, the device 10 may repeat the processes by the causal model acquisition unit 11, the importance calculation unit 12, and the reconstruction unit 13 using the reconstructed model instead of the causal model acquired in S1 until the number of candidate variables becomes equal to or less than a predetermined number. The predetermined number is not particularly limited, and may be set to any number, for example. The predetermined number may be, for example, 1 or a multiple number of 2 or more, but is preferably 1. The predetermined number may be, for example, a ratio to the number of variables included in the causal model acquired in S1. The ratio is not particularly limited, and may be, for example, 50%, 25%, 15%, 10%, etc.

[0030] Then, the result output unit 14 outputs at least one of the reconstructed models (S4, result output step). The result output unit 14 may output the reconstructed model to the output device 106 (e.g., a display) of the apparatus 10, or to an external device. The external device is not particularly limited, and examples thereof include a printer and a terminal device (e.g., a user's smartphone, PC, tablet terminal, etc.).

[0031] According to the causal analysis support device of this embodiment, the importance calculation unit can calculate the importance of the variables included in the causal model acquired by the causal model acquisition unit. Then, the reconstruction unit generates at least one reconstructed model by reconstructing the causal model based on the importance, and the result output unit can output at least one reconstructed model. Therefore, according to the causal analysis support device of this embodiment, a reconstructed model that takes into account the importance of the variables included in the causal model can be generated, thereby suppressing a decrease in the degree of conformance and improving the readability of the causal model. Therefore, the causal analysis support device of this embodiment can be used in fields that use analysis of psychological causality, and can be particularly suitable for estimating causal relationships between a large number of variables in the fields of IT, economics, life science, environment, sociology, etc.

[0032] [Example 1] It has been confirmed that the causal analysis support device of the present disclosure is suitable for causal analysis in multivariate data with a large number of causal effects. Note that the present disclosure is not limited to the following examples.

[0033] First, data on 22 variables (variable Q, variable C, variable I, variable K, variable G, variable R, variable O, variable D, variable L, variable T, variable J, variable U, variable B, variable N, variable V, variable A, variable P, variable S, variable M, variable E, variable H, variable F) was prepared, and a causal model for evaluation was prepared using LiNGAM. Figure 6(A) shows an overall view of the causal model for evaluation. As shown in Fig. 6(B), the goodness of fit of the evaluation causal model was CFI (Comparative Fit Index) 0.990717, GFI (Goodness of Fit Index) 0.977986, AGFI (Adjusted Goodness of Fit Index) 0.960845, RMSEA (Root Mean Square Error of Approximation) 0.026482, AIC (Akaike's Information Criterion) 243.554451, and BIC (Bayesian Information Criterion) 842.300595. As shown in Fig. 6(B), the evaluation causal model in Fig. 6(A) has a high goodness of fit and is considered to be a valid model for the causal relationships between the 22 variables, but as shown in Fig. 6(A), the causal relationships between the variables are very complicated and difficult to read.

[0034] FIG. 7 shows an example of a reconstructed model obtained by inputting the evaluation causal model of FIG. 6(A) into the present device 10 and reconstructing it. As shown in FIG. 7(A), the reconstructed model reconstructed by the present device 10 has reduced variables and improved readability. As shown in FIG. 7(B), the reconstructed model has a CFI (Comparative Fit Index) of 0.97844, a GFI (Goodness of Fit Index) of 0.95897, an AGFI (Adjusted Goodness of Fit Index) of 0.92740, an RMSEA (Root Mean Square Error of Approximation) of 0.03206, an AIC (Akaike's Information Criterion) of 29.94730, and a BIC (Bayesian Information Criterion) of 103.56363. As shown in FIGS. 7(A) and (B), the reconstructed model reconstructed by the present device 10 can suppress a decrease in the goodness of fit of the model and reduce variables. For this reason, it has been found that the present device 10 can suppress a decrease in the goodness of fit of the causal model and improve the readability of the causal model.

[0035] A possible method for simply improving the readability of a causal model is, for example, to extract and interpret relationships around the target variable to be analyzed. As a reference example, for the evaluation causal model in Figure 6(A), an example of a reference causal model reconstructed using the picked variables is shown in Figure 8. As shown in Figure 8(A), the readability of the causal model is improved by the method of the reference example, but as shown in Figure 8(B), the causal model of the reference example had a CFI (Comparative Fit Index) of 0.915675, a GFI (Goodness of Fit Index) of 0.897672, an AGFI (Adjusted Goodness of Fit Index) of 0.870684, an RMSEA (Root Mean Square Error of Approximation) of 0.063892, an AIC (Akaike's Information Criterion) of 57.075787, and a BIC (Bayesian Information Criterion) of 199.400690. The causal model of the reference example had a significant decrease in model fit, with the values ​​of CFI, GFI, AGFI, and RMSEA decreasing from the evaluation causal model in Figure 6, and it was found that the decrease in GFI and AGFI in particular was at an unacceptable level.

[0036] [Embodiment 2] The second embodiment is an example of a causal analysis support device according to the present disclosure.

[0037] FIG. 9 is a block diagram showing an example of the configuration of the causal analysis support device 10A. As shown in FIG. 9, the causal analysis support device 10A includes an evaluation value calculation unit 15 in addition to the configuration of the causal analysis support device 10 of the first embodiment. The hardware configuration of the causal analysis support device 10A is the same as that of the causal analysis support device 10 of FIG. 2 except that the central processing unit 101 has the configuration of the causal analysis support device 10A of FIG. 9 instead of the configuration of the causal analysis support device 10 of FIG. 1. The process of the evaluation value calculation unit 15 will be described below. The process of the evaluation value calculation unit 15 can be inserted at any position in the flowchart of FIG. 3 described in the first embodiment, for example, but it is preferable to insert it after S3 and before S4 as shown in FIG. 10.

[0038] The evaluation value calculation unit 15 calculates at least one evaluation value of the causal model and the reconstructed model (S11, evaluation value calculation step). The evaluation value is, for example, an index that comprehensively evaluates the fitness and readability of the causal model and the reconstructed model. The evaluation value calculation unit 15 can calculate the evaluation value using an arbitrary score function based on the fitness and number of variables of each model. The score function is not particularly limited, and an example of the score function is a function such as the following formula (1). Note that the score function in the following formula (1) is an example, and the present disclosure is not limited to the following example. The evaluation value calculation unit 15 may, for example, store the calculated evaluation value in the storage unit in association with the corresponding causal model or reconstructed model. Equation (1): (0.97-CFI)^2 + 2*(0.04-RMSEA)^2 +0.3*(3-number of variables)^2

[0039] In this embodiment, the result output unit 14 may, for example, in the result output step of S4, select at least one of the reconstructed model and the causal model based on the evaluation value, and output the selected reconstructed model and the causal model. The result output unit 14 may, for example, select the reconstructed model or the causal model having the best evaluation value, or may select at least one reconstructed model or the causal model having the evaluation value exceeding the evaluation criterion. The selection criterion of the reconstructed model by the result output unit 14 is not particularly limited, and for example, an appropriate criterion can be set according to the calculation method of the evaluation value.

[0040] According to the causal analysis support device of this embodiment, for the causal model acquired by the causal model acquisition unit, the importance calculation unit can calculate the importance of the variables included in the causal model. Then, the reconstruction unit generates at least one reconstructed model by reconstructing the causal model based on the importance, the evaluation value calculation unit calculates at least one evaluation value of the causal model and the reconstructed model, and the result output unit selects at least one of the reconstructed model and the causal model based on the evaluation value, and outputs the selected reconstructed model and the causal model. Therefore, according to the causal analysis support device of this embodiment, a reconstructed model that takes into account the importance of the variables included in the causal model can be generated, so that the deterioration of the degree of conformance can be suppressed and the readability of the causal model can be improved. Furthermore, according to the causal analysis support device of this embodiment, the result output unit can select at least one of the reconstructed model and the causal model based on the evaluation value calculated by the evaluation value calculation unit, and output the selected reconstructed model and the causal model. Therefore, according to the causal analysis support device of this embodiment, the user can more easily view the causal model in which the decrease in the degree of conformance is suppressed and the readability is improved. Therefore, the causal analysis support device of this embodiment can be used in fields that use the analysis of psychological causality, and can be particularly suitably used for estimating the causal relationships between a large number of variables in the fields of IT, economics, life science, environment, sociology, etc.

[0041] [Embodiment 3] The program of this embodiment is a program for causing a computer to execute each step of the above-mentioned causal analysis support method. Specifically, the program of this embodiment is a program for causing a computer to execute a causal model acquisition procedure, a importance calculation procedure, a reconstruction procedure, and a result output procedure.

[0042] The causal model acquisition step acquires a causal model including a causal relationship of at least one variable; The importance calculation step calculates the importance of the variables included in the causal model; The reconstruction step reconstructs the causal model based on the importance to generate at least one reconstructed model; The result output step outputs at least one of the reconstructed models.

[0043] The program of the present embodiment can also be said to be a program that causes a computer to function as a causal model acquisition procedure, an importance calculation procedure, a reconstruction procedure, and a result output procedure.

[0044] The program of this embodiment can be implemented by using the description of the causal analysis support device and the causal analysis support method of the present disclosure. For example, the "procedure" in each of the procedures can be replaced with "processing". The program of this embodiment may be recorded in a computer-readable recording medium. The recording medium is, for example, a non-transitory computer-readable storage medium. The recording medium is not particularly limited, and examples thereof include random access memory (RAM), read-only memory (ROM), hard disk (HD), flash memory (for example, SSD (Solid State Drive), USB flash memory, SD / SDHC card, etc.), optical disk (for example, CD-R / CD-RW, DVD-R / DVD-RW, BD-R / BD-RE, etc.), magneto-optical disk (MO), floppy (registered trademark) disk (FD), etc. The program of this embodiment (for example, also referred to as a programming product or program product) may be distributed from an external computer, for example. The "distribution" may be, for example, distribution via a communication network or distribution via a device connected by a wire. The program of the present embodiment may be installed and executed in the device to which it is distributed, or may be executed without being installed.

[0045] Although the present disclosure has been described above with reference to the embodiments, the present disclosure is not limited to the above-mentioned embodiments. Various modifications that can be understood by a person skilled in the art can be made to the configuration and details of the present disclosure within the scope of the present disclosure. Furthermore, each embodiment can be appropriately combined with other embodiments.

[0046] <Additional Notes> Some or all of the above embodiments may be described as follows, but are not limited to the following supplementary notes. (Appendix 1) The present invention includes a causal model acquisition unit, an importance calculation unit, a reconstruction unit, and a result output unit, the causal model acquisition unit acquires a causal model including a causal relationship of at least one variable; The importance calculation unit calculates the importance of the variables included in the causal model; The reconstructing unit reconstructs the causal model based on the importance to generate at least one reconstructed model; The result output unit outputs at least one of the reconstructed models. (Appendix 2) 2. The causal analysis support device according to claim 1, wherein the importance calculation unit calculates the importance based on a causal coefficient of the variable. (Appendix 3) 3. The causal analysis support device according to claim 2, wherein the importance calculation unit calculates the importance based on an absolute value of the causal coefficient. (Appendix 4) 4. The causal analysis support device according to claim 1, wherein the importance calculation unit calculates a centrality index of the variable as the importance. (Appendix 5) The reconstruction unit is generating candidate variables by deleting at least one variable from the variables included in the causal model based on the importance; A causal analysis support device according to any one of appendix 1 to 4, which generates at least one reconstructed model based on the candidate variables. (Appendix 6) The causal analysis support device according to claim 5, wherein the causal model generation unit generates at least one reconstructed model with the target variable as an end point. (Appendix 7) 7. The causal analysis support device according to claim 5, wherein the causal model generation unit generates at least one reconstructed model starting from a target variable. (Appendix 8) An evaluation value calculation unit is included, 8. The causal analysis support device according to claim 1, wherein the evaluation value calculation unit calculates an evaluation value of at least one of the causal model and the reconstructed model. (Appendix 9) The causal analysis support device according to claim 8, wherein the result output unit selects at least one of the reconstructed model and the causal model based on the evaluation value, and outputs the selected reconstructed model and the causal model. (Appendix 10) The method includes a causal model acquisition step, a importance calculation step, a reconstruction step, and a result output step, The causal model acquisition step acquires a causal model including a causal relationship of at least one variable; The importance calculation step calculates the importance of the variables included in the causal model, The reconstructing step reconstructs the causal model based on the importance to generate at least one reconstructed model; The causal analysis support method, wherein the result output step outputs at least one of the reconstructed models. (Appendix 11) The causal analysis support method according to claim 10, wherein the importance calculation step calculates the importance based on a causal coefficient of the variable. (Appendix 12) 12. The causal analysis support method according to claim 11, wherein the importance calculation step calculates the importance based on an absolute value of the causal coefficient. (Appendix 13) 13. The causal analysis support method according to any one of appendices 10 to 12, wherein the importance calculation step calculates a centrality index of the variable as the importance. (Appendix 14) The reconstructing step comprises: generating candidate variables by deleting at least one variable from the variables included in the causal model based on the importance; A causal analysis support method according to any one of appendices 10 to 13, further comprising generating at least one reconstructed model based on the candidate variables. (Appendix 15) The causal analysis support method according to claim 14, wherein the causal model generation step generates at least one reconstructed model with the target variable as an end point. (Appendix 16) 16. The causal analysis support method according to claim 14 or 15, wherein the causal model generation step generates at least one reconstructed model starting from a target variable. (Appendix 17) An evaluation value calculation step is included, The causal analysis support method according to any one of appendices 10 to 16, wherein the evaluation value calculation step calculates an evaluation value of at least one of the causal model and the reconstructed model. (Appendix 18) The causal analysis support method according to claim 17, wherein the result output step selects at least one of the reconstructed model and the causal model based on the evaluation value, and outputs the selected reconstructed model and the causal model. (Appendix 19) It includes a procedure for obtaining a causal model, a procedure for calculating importance, a procedure for reconstructing the model, and a procedure for outputting the results. The causal model acquisition step acquires a causal model including a causal relationship of at least one variable; The importance calculation step calculates the importance of the variables included in the causal model; The reconstruction step reconstructs the causal model based on the importance to generate at least one reconstructed model; The result output step outputs at least one of the reconstructed models; A program for causing a computer to execute each of the above procedures. (Appendix 20) 20. The program according to claim 19, wherein the importance calculation step calculates the importance based on a causal coefficient of the variable. (Appendix 21) 21. The program according to claim 20, wherein the importance calculation step calculates the importance based on an absolute value of the causal coefficient. (Appendix 22) 22. The program according to any one of appendices 19 to 21, wherein the importance calculation step calculates a centrality index of the variable as the importance. (Appendix 23) The reconstruction procedure comprises: generating candidate variables by deleting at least one variable from the variables included in the causal model based on the importance; 23. The program of any one of claims 19 to 22, further comprising generating at least one reconstruction model based on the candidate variables. (Appendix 24) The program of claim 23, wherein the causal model generation step generates at least one reconstructed model with the target variable as an end point. (Appendix 25) 25. The program according to claim 23 or 24, wherein the causal model generation procedure generates at least one reconstructed model starting from a target variable. (Appendix 26) Including an evaluation value calculation procedure, 26. The program described in any one of Appendixes 19 to 25, wherein the evaluation value calculation step calculates an evaluation value of at least one of the causal model and the reconstructed model. (Appendix 27) The program described in Appendix 26, wherein the result output step selects at least one of the reconstructed model and the causal model based on the evaluation value, and outputs the selected reconstructed model and the causal model. (Appendix 28) It includes a procedure for obtaining a causal model, a procedure for calculating importance, a procedure for reconstructing the model, and a procedure for outputting the results. The causal model acquisition step acquires a causal model including a causal relationship of at least one variable; The importance calculation step calculates the importance of the variables included in the causal model; The reconstruction step reconstructs the causal model based on the importance to generate at least one reconstructed model; The result output step outputs at least one of the reconstructed models; A computer-readable recording medium having recorded thereon a program for causing a computer to execute each of the above procedures. (Appendix 29) 29. The recording medium according to claim 28, wherein the importance calculation step calculates the importance based on a causal coefficient of the variable. (Appendix 30) 21. The recording medium according to claim 20, wherein the importance calculation step calculates the importance based on an absolute value of the causal coefficient. (Appendix 31) 31. The recording medium according to any one of appendices 28 to 30, wherein the importance calculation step calculates a centrality index of the variable as the importance. (Appendix 32) The reconstruction procedure comprises: generating candidate variables by deleting at least one variable from the variables included in the causal model based on the importance; 32. A recording medium according to any one of claims 28 to 31, further comprising: generating at least one reconstruction model based on the candidate variables. (Appendix 33) 33. The recording medium of claim 32, wherein the causal model generation procedure generates at least one reconstructed model that has a target variable as an end point. (Appendix 34) 34. The recording medium of claim 32 or 33, wherein the causal model generation procedure generates at least one reconstructed model starting from a target variable. (Appendix 35) Including an evaluation value calculation procedure, A recording medium described in any one of Addendums 28 to 34, wherein the evaluation value calculation step calculates an evaluation value of at least one of the causal model and the reconstructed model. (Appendix 36) The recording medium described in Appendix 35, wherein the result output step selects at least one of the reconstructed model and the causal model based on the evaluation value, and outputs the selected reconstructed model and the causal model. [Industrial Applicability]

[0047] For this reason, the present disclosure can be used, for example, in fields that utilize causal analysis, and is particularly suitable for estimating causal relationships between a large number of variables in fields such as IT, economics, life sciences, the environment, and sociology. [Explanation of symbols]

[0048] 10, 10A Causal analysis support device 11 Causal model acquisition section 12 Importance calculation part 13 Reconstruction part 14 Result output section 15 Evaluation value calculation section 101 CPU 102 Memory 103 Bus 104 Storage device 105 Input Device 106 Output Device 107 Communication Devices

Claims

1. The present invention includes a causal model acquisition unit, an importance calculation unit, a reconstruction unit, and a result output unit, the causal model acquisition unit acquires a causal model including a causal relationship of at least one variable; The importance calculation unit calculates the importance of the variables included in the causal model; The reconstructing unit reconstructs the causal model based on the importance to generate at least one reconstructed model; The result output unit outputs at least one of the reconstructed models.

2. The causal analysis support device according to claim 1 , wherein the importance calculation unit calculates the importance based on a causal coefficient of the variable.

3. The causal analysis support device according to claim 2 , wherein the importance calculation unit calculates the importance based on an absolute value of the causal coefficient.

4. The causal analysis support device according to claim 1 , wherein the importance calculation unit calculates a centrality index of the variable as the importance.

5. The reconstruction unit is generating candidate variables by deleting at least one variable from the variables included in the causal model based on the importance; The causal analysis support device according to claim 1 , further comprising: generating at least one reconstructed model based on the candidate variables.

6. An evaluation value calculation unit is included, The causal analysis support device according to claim 1 , wherein the evaluation value calculation unit calculates an evaluation value of at least one of the causal model and the reconstructed model.

7. The causal analysis support device according to claim 6 , wherein the result output unit selects at least one of the reconstructed model and the causal model based on the evaluation value, and outputs the selected reconstructed model and the causal model.

8. The method includes a causal model acquisition step, a importance calculation step, a reconstruction step, and a result output step, The causal model acquisition step acquires a causal model including a causal relationship of at least one variable; The importance calculation step calculates the importance of the variables included in the causal model, The reconstructing step reconstructs the causal model based on the importance to generate at least one reconstructed model; The causal analysis support method, wherein the result output step outputs at least one of the reconstructed models.

9. It includes a procedure for obtaining a causal model, a procedure for calculating importance, a procedure for reconstructing the model, and a procedure for outputting the results. The causal model acquisition step acquires a causal model including a causal relationship of at least one variable; The importance calculation step calculates the importance of the variables included in the causal model; The reconstruction step reconstructs the causal model based on the importance to generate at least one reconstructed model; The result output step outputs at least one of the reconstructed models; A program for causing a computer to execute each of the above procedures.

10. It includes a procedure for obtaining a causal model, a procedure for calculating importance, a procedure for reconstructing the model, and a procedure for outputting the results. The causal model acquisition step acquires a causal model including a causal relationship of at least one variable; The importance calculation step calculates the importance of the variables included in the causal model; The reconstruction step reconstructs the causal model based on the importance to generate at least one reconstructed model; The result output step outputs at least one of the reconstructed models; A computer-readable recording medium having recorded thereon a program for causing a computer to execute each of the above procedures.

Citation Information

Patent Citations

  • Analysis device, analysis method, and analysis program

    JP2020149303A