JP7704229B2 - RULE GENERATION DEVICE, RULE GENERATION METHOD, AND PROGRAM - Google Patents

Description

The present invention relates to a rule generation device, a judgment device, a rule generation method, a judgment method, and a program.

In fields such as economics, where many different factors are interrelated, when trying to achieve a certain goal, it is necessary to determine what approach will be effective.

For example, Patent Document 1 discloses a system that creates a model formula using actual values of multiple economic indicators and performs an econometric model simulation.

Furthermore, Patent Document 2 describes predicting future macroeconomic indicators, future management indicators of individual companies, future stock prices of individual companies, etc. based on exogenous macroeconomic indicators.

Japanese Patent Application Publication No. 6-231108 Japanese Patent Application Publication No. 10-3465

However, the above-mentioned Patent Documents 1 and 2 only make future predictions based on assumptions, which creates the problem that it is not possible to clarify which approach would be effective in achieving the desired goal.

One example of the object of the present invention is to provide a rule generation device, a judgment device, a rule generation method, a judgment method, and a program that solve the above-mentioned problems.

According to one aspect of the present invention,
An acquisition means for acquiring time series data of a plurality of indices and condition information indicating a condition related to a target index;
a characteristic indicator specifying means for specifying at least one of the plurality of indicators as a characteristic indicator for the target indicator by using time-series data of the plurality of indicators and the condition information;
a generating means for acquiring values of one or more of the characteristic indicators, and generating, using at least the values, one or more rules that are estimated to be necessary for achieving a condition related to the objective indicator;
A rule generating device is provided, in which the rule includes at least one sub-rule related to the feature indicator.

According to one aspect of the present invention,
A first acquisition means for acquiring one or more index values and condition information indicating a condition for a target index;
a second acquisition means for acquiring one or more rules, each of which includes one or more subrules related to the index, in response to the condition information;
and determining means for determining whether or not the acquired values of the one or more indices satisfy at least any of the one or more subrules.

According to one aspect of the present invention,
One or more computers
Acquire time series data of a plurality of indices and condition information indicating a condition related to a target index;
Using the time-series data of the plurality of indices and the condition information, at least one of the plurality of indices is identified as a characteristic indices for the target indices;
obtaining values of one or more of the characteristic indicators, and using at least the values to generate one or more rules that are estimated to be necessary to achieve a condition related to the objective indicator;
A rule generating method is provided in which the rule includes at least one sub-rule related to the feature indicator.

According to one aspect of the present invention,
One or more computers
Obtaining values of one or more indicators and condition information indicating a condition for the objective indicator;
Obtaining one or more rules, each including one or more subrules related to the indicator, in response to the condition information;
A method for determining whether the obtained values of the one or more indices satisfy at least any of the one or more sub-rules is provided.

According to one aspect of the present invention,
Computer,
An acquisition means for acquiring time series data of a plurality of indices and condition information indicating a condition related to the target index;
a characteristic indicator specifying means for specifying at least one of the plurality of indicators as a characteristic indicator for the target indicator using the time-series data of the plurality of indicators and the condition information; and a generating means for acquiring values of one or more of the characteristic indicators and generating one or more rules estimated to be necessary for achieving a condition related to the target indicator using at least the values.
A program is provided in which the rules include at least one sub-rule related to the characteristic indicator.

According to one aspect of the present invention,
Computer,
A first acquisition means for acquiring one or more index values and condition information indicating a condition for the target index;
There is provided a second acquisition means for acquiring one or more rules, each of which includes one or more subrules related to the index, in accordance with the condition information, and a program that functions as a determination means for determining whether or not the acquired values of the one or more indexes satisfy at least any of the one or more subrules.

According to one aspect of the present invention, a rule generation device, a judgment device, a rule generation method, a judgment method, and a program are provided that can clarify the conditions necessary to achieve a goal.

1 is a diagram showing an overview of a rule generating device according to a first embodiment; 1 is a block diagram illustrating a functional configuration of a rule generating device according to a first embodiment. 10 is a diagram illustrating an example of time-series data of multiple indexes acquired by an acquisition unit. 11 is a diagram for explaining a process performed by a feature index identifying unit; FIG. FIG. 11 is a diagram showing a specific example of information output from a first model. 11 is a diagram illustrating examples of values of one or more feature indices acquired by a generating unit. FIG. FIG. 13 is a diagram illustrating an example of a configuration of output data of a second model. FIG. 11 is a diagram showing a specific example of an output rule output from a generation unit. 11 is a diagram illustrating an example of a configuration of an output rule output from a generating unit. FIG. 11 is a diagram illustrating an example of a configuration of an output rule output from a generating unit. FIG. 11 is a diagram illustrating an example of a configuration of an output rule output from a generating unit. FIG. FIG. 13 is a diagram illustrating a computer for realizing the rule generating device. 1 is a flowchart illustrating the flow of a rule generation method according to the first embodiment. FIG. 13 is a diagram illustrating an overview of a determination device according to a second embodiment. FIG. 11 is a diagram illustrating a functional configuration of a determination device according to a second embodiment. FIG. 11 is a diagram illustrating an example of rule set reference information. 10 is a flowchart illustrating the flow of a determination method according to a second embodiment. FIG. 13 is a block diagram illustrating a functional configuration of a determination device according to a third embodiment. FIG. 13 is a diagram for explaining a determination device according to a fourth embodiment.

Hereinafter, an embodiment of the present invention will be described with reference to the drawings. In all drawings, similar components are given similar reference symbols and descriptions will be omitted as appropriate.

(First embodiment)
FIG. 1 is a diagram showing an overview of a rule generating device 10 according to the first embodiment. The rule generating device 10 according to this embodiment includes an acquisition unit 120, a feature index identification unit 140, and a generation unit 160. The acquisition unit 120 acquires time-series data of a plurality of indexes and condition information indicating a condition related to a target index. The feature index identification unit 140 uses the time-series data of the plurality of indexes and the condition information to identify at least one of the plurality of indexes as a feature index for the target index. The generation unit 160 acquires values of one or more feature indexes, and generates one or more rules estimated to be necessary to achieve a condition related to the target index using at least the values. The rule includes at least one sub-rule related to the feature index.

This rule generation device 10 makes it possible to clarify the conditions necessary to achieve a goal.

Below, a detailed example of the rule generation device 10 is described.

Each of the multiple indicators and the objective indicator is, for example, an economic indicator. There are tens of thousands of types of economic indicators, and there are many indicators for which time series data can be obtained as open data. However, each of the multiple indicators and the objective indicator is not particularly limited, and may be an indicator indicating a social situation, an indicator indicating a physical condition, etc. Below, an example in which the multiple indicators and the objective indicator are economic indicators is described.

For example, suppose a certain measure is implemented to achieve the goal of improving a certain target index. However, the target index may not necessarily improve as intended. The reason for this is that there are other factors that increase the target index, and those factors do not meet the sufficient conditions. There is not necessarily just one such factor. It is believed that the goal can be achieved by clarifying the factors and taking an appropriate approach. However, there are tens of thousands of economic indicators. It is necessary to identify the factors from among these indicators that are related to the target index. Furthermore, by clarifying the specific conditions that the identified indicator must meet, the measures aimed at achieving the goal become clearer, and the effectiveness of the measures will be increased.

Figure 2 is a block diagram illustrating the functional configuration of the rule generation device 10 according to this embodiment. In the example of this figure, the rule generation device 10 further includes an index storage unit 122.

<Acquisition unit 120>
3 is a diagram illustrating an example of time series data of multiple indexes acquired by the acquisition unit 120. The time series data of multiple indexes (index 1, index 2, ..., index N) is stored in advance in the index storage unit 122. The acquisition unit 120 can read and acquire the time series data from the index storage unit 122. In the example of FIG. 2, the index storage unit 122 is provided in the rule generation device 10, but the index storage unit 122 may be provided outside the rule generation device 10. Moreover, the acquisition unit 120 may acquire the time series data from another device instead of reading it from the index storage unit 122. The number N of the multiple indexes is not particularly limited. The number of the multiple indexes may be 10,000 or more.

The acquisition unit 120 also acquires condition information. The condition information is, for example, information indicating the value of the target index or the range of the target index. The target index is one of a plurality of indexes (index 1, index 2, ..., index N). The condition information is, for example, input to the rule generation device 10 by a user, and the acquisition unit 120 can acquire the input condition information. Alternatively, the condition information may be stored in advance in a storage unit accessible from the acquisition unit 120, and the acquisition unit 120 may read and acquire it, or the acquisition unit 120 may acquire the condition information from an external device.

<Feature Index Identification Unit 140>
FIG. 4 is a diagram for explaining the process performed by the feature index identifying unit 140. The feature index identifying unit 140 extracts one or more of the multiple indices and the change rates that represent the transition of the multiple indices over time, which affect the target index. The feature index identifying unit 140 then identifies the feature index based on at least one of the extracted indices and the change rates. Specifically, the feature index identifying unit 140 includes a first model 142 for extracting feature amounts. The feature index identifying unit 140 then uses the first model 142 to extract one or more of the multiple indices and the change rates that affect the target index. By identifying the feature index in this way, the feature index identifying unit 140 can narrow down the multiple indices to those that contribute greatly to achieving the conditions related to the target index, and generate rules.

In the example shown in this figure, the time-series data and condition information acquired by the acquisition unit 120 are input to the feature index identification unit 140, which then inputs them to the first model 142. The first model 142 then outputs information indicating one or more of index 1, index 2, ..., index N, the rate of change of index 1, the rate of change of index 2, ..., and the rate of change of index N. What is output from the first model 142 is the index or rate of change that has the greatest impact on the target index among the input indices.

The feature index identification unit 140 takes the index shown in the output data of the first model 142 as the feature index. The feature index identification unit 140 also takes the index that is the source of the change rate shown in the output data of the first model 142 as the feature index. That is, if the "change rate of index n" is shown in the output data of the first model 142, the feature index identification unit 140 takes "index n" as the feature index. In the example of this figure, specifically, the output data of the first model 142 shows the change rate of index 2, the change rate of index 15, the change rate of index 82, index 190, ..., and index 8734. The feature index identification unit 140 then takes index 2, index 15, index 82, index 190, ..., and index 8734 as feature index 1, feature index 2, feature index 3, feature index 4, ..., and feature index M, respectively.

The characteristic index identification unit 140 does not need to identify characteristic indexes based on all the indexes and change rates extracted by the first model 142. For example, the remaining indexes, excluding those that are difficult to approach by measures, among the indexes shown in the output data of the first model 142 and the indexes that are the basis of the change rates shown in the output data, may be used as characteristic indexes. In such a case, it is sufficient to predetermine a group of indexes, among index 1, index 2, ..., index N, that are not to be used as characteristic indexes, or a group of indexes that can be used as characteristic indexes. Then, the characteristic index identification unit 140 can identify characteristic indexes according to the determined group of indexes.

The number M of feature indices identified by the feature index identification unit 140 is not particularly limited. However, the number M of feature indices is smaller than the number N of the multiple indices input to the first model 142. M may be, for example, 1/10 or less of N, or 1/100 or less. M is, for example, 100 or less. M may be, for example, 10 or more, or 20 or more.

For example, the feature index identifying unit 140 may identify a predetermined number Mp of feature indexes based on the extraction results by the first model 142. In this case, the extraction results by the first model 142 are ranked to indicate the strength of correlation with the condition information, and the feature index identifying unit 140 can determine the feature index based on Mp highly correlated indexes or change rates.

The first model 142 can be realized using existing technology. The first model 142 is a trained model based on machine learning. This machine learning can be performed using time series data of a combination of the value of the objective index, other index values, and a correct/incorrect flag as training data. The correct/incorrect flag can be assigned based on a specified condition for the objective index. Here, the specified condition may be the same as the condition indicated in the condition information, or it may not be the same.

In this figure, an example is shown in which the first model 142 extracts one or more of the multiple indicators and the change rates representing the time trends of the multiple indicators that affect the target indicator. As shown in this figure, it is preferable for the first model 142 to extract at least one change rate of the indicator. By extracting not only the indicator itself but also candidates including the change rate of the indicator, analysis taking into account the time trends of each indicator can be performed. Therefore, the rule generation device 10 can generate more accurate rules. Note that even in this case, the acquisition unit 120 does not need to acquire time series data of the change rate of the indicator separately from the time series data of the indicator. However, the first model 142 may extract one or more of the multiple indicators that affect the target indicator without considering the change rate.

Figure 5 is a diagram showing a specific example of information output from the first model 142. In this figure, the numbers at the beginning of each line are identification codes for each index. In addition to the index and information indicating the rate of change of the index, the output data of the first model 142 may also indicate situations that are highly correlated with situations that satisfy the condition information input to the first model 142, as shown in this figure. Even in this case, it is useful to generate more specific rules by the processing of the generation unit 160, which will be described later.

<Generation unit 160>
As described above, the generation unit 160 obtains values of one or more characteristic indicators, and uses at least the values to generate one or more rules that are estimated to be necessary to achieve a condition related to the objective indicator.

Figure 6 is a diagram illustrating an example of values of one or more feature indicators acquired by the generation unit 160. The values acquired by the generation unit 160 may be, for example, values of indicators corresponding to each feature indicator among the time series data of multiple indicators acquired by the acquisition unit 120. Furthermore, the values acquired by the generation unit 160 may be, for example, values indicating the latest situation among the time series data. Additionally, the generation unit 160 may accept input of feature indicator values by the user separately from the time series data acquired by the acquisition unit 120.

The number of feature index values acquired by the generation unit 160 is not particularly limited, and may be, for example, 100 or less, or 50 or less. The number of feature index values acquired by the generation unit 160 may be, for example, 10 or more, or 20 or more.

The generation unit 160 includes a second model 162 based on ensemble machine learning. The second model 162 is, for example, a trained model based on Ensemble Machine Learning (EML), which has high interpretability. This machine learning can be performed using time-series data of a combination of the value of the objective index, other index values, and a correct/incorrect flag as training data. The correct/incorrect flag can be assigned based on a specified condition for the objective index. Here, the specified condition may be the same as the condition indicated in the condition information, or it may not be the same.

Figure 7 is a diagram illustrating the configuration of output data of the second model 162. In the generation unit 160, the value of the feature index and condition information are input to the second model 162. As the value of the feature index, one value may be input for each feature index. Furthermore, the number of feature indexes whose values are input to the second model 162 is not particularly limited. Then, multiple rules are output from the second model 162. Each rule includes one or more subrules. Each subrule indicates a condition related to one of the feature indexes. Here, the subrules output from the second model 162 may include conditions that are satisfied by the value of the feature index input to the second model 162 and conditions that are not satisfied.

The number of subrules included in each rule can be preset in the second model 162. When multiple rules are generated from the second model 162, the numbers of subrules included in the rules may be different from each other or may be the same.

Figure 8 is a diagram showing a specific example of an output rule output from the generation unit 160. In this figure, the number at the beginning of each subrule is the identification code of each index. If the rule output from the second model 162 includes a rule that includes multiple subrules, the generation unit 160 connects those subrules with "AND". That is, for each rule, the rule being satisfied means that all subrules included in the rule are satisfied. Also, if multiple rules are output from the second model 162, the generation unit 160 connects those rules with "AND" or "OR". In this way, the generation unit 160 outputs one generated rule or two or more rules connected to each other as an output rule. The output rule output from the generation unit 160 may be displayed on a display, for example, or may be stored in a storage device accessible from the generation unit 160. The user can grasp the output rule output from the generation unit 160 as a condition necessary to satisfy the condition related to the target index indicated in the condition information.

In the example shown in this figure, the generation unit 160 outputs the rule obtained by connecting rule 1 and rule 2 with "AND" as the final output rule. In rule 1 and rule 2, the sub-rules contained in each rule are connected to each other with "AND".

Figure 9 is a diagram showing an example of the configuration of an output rule output from the generation unit 160. The one or more rules generated by the second model 162 in the generation unit 160 may include two or more rules that are mutually interchangeable. In the example shown in this figure, rule 2-1 and rule 2-2 are mutually interchangeable, and it is sufficient if at least one of them is satisfied. By generating alternative rules in this manner, even if it is difficult to satisfy any of the rules, an alternative measure can be implemented.

Figures 10 and 11 are diagrams showing an example of the configuration of the output rules output from the generation unit 160. The number of rules output from the second model 162 is not particularly limited, and can be set in advance in the second model 162. The generation unit 160 can generate output rules of a predetermined structure using the output data of the second model 162. For example, in the example of Figure 10, in the output rules, the rule in which rules 2-2-1 and 2-2-2 are connected with "AND" can be replaced with rule 2-1. In the example of Figure 11, in the output rules, three or more rules are connected with "AND". Note that the configurations of the output rules shown in Figures 8 to 11 are examples, and are not limited to these.

In addition, the output rules output from the generation unit 160 may include all of the rules output from the second model 162, or may include only some of them. In addition, the output rules output from the generation unit 160 may include sub-rules for all feature indicators, or may include sub-rules for only some of the feature indicators.

The hardware configuration of the rule generation device 10 is described below. Each functional component of the rule generation device 10 is realized by a combination of hardware and software (e.g., a combination of an electronic circuit and a program that controls it). This is further described below.

Figure 12 is a diagram illustrating a computer 1000 for realizing the rule generation device 10. The rule generation device 10 may be realized by one computer 1000, or may be realized by the cooperation of multiple computers 1000. The computer 1000 is any computer. For example, the computer 1000 is a SoC (System On Chip), a Personal Computer (PC), a server machine, a tablet terminal, or a smartphone. The computer 1000 may be a dedicated computer designed to realize the rule generation device 10, or may be a general-purpose computer.

The computer 1000 has a bus 1020, a processor 1040, a memory 1060, a storage device 1080, an input/output interface 1100, and a network interface 1120. The bus 1020 is a data transmission path for the processor 1040, the memory 1060, the storage device 1080, the input/output interface 1100, and the network interface 1120 to transmit and receive data to each other. However, the method of connecting the processor 1040 and the like to each other is not limited to bus connection. The processor 1040 is various processors such as a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), or an FPGA (Field-Programmable Gate Array). The memory 1060 is a main storage device realized using a RAM (Random Access Memory) or the like. The storage device 1080 is an auxiliary storage device realized using a hard disk, an SSD (Solid State Drive), a memory card, or a ROM (Read Only Memory), or the like.

The input/output interface 1100 is an interface for connecting the computer 1000 to an input/output device. For example, the input/output interface 1100 is connected to an input device such as a keyboard and an output device such as a display.

The network interface 1120 is an interface for connecting the computer 1000 to a network. This communication network is, for example, a LAN (Local Area Network) or a WAN (Wide Area Network). The method by which the network interface 1120 connects to the network may be a wireless connection or a wired connection.

The storage device 1080 stores program modules that realize each functional component of the rule generation device 10. The processor 1040 reads each of these program modules into the memory 1060 and executes them to realize the function corresponding to each program module.

In addition, when the index memory unit 122 is provided inside the rule generation device 10, for example, the index memory unit 122 is realized using a storage device 1080.

Figure 13 is a flowchart illustrating the flow of the rule generation method according to this embodiment. The rule generation method according to the embodiment is executed by one or more computers. The rule generation method according to this embodiment includes steps S110, S120, and S130. In step S110, time series data of multiple indicators and condition information indicating a condition related to the target indicator are acquired. In step S120, at least one of the multiple indicators is identified as a feature indicator for the target indicator using the time series data of the multiple indicators and the condition information. In step S130, values of one or more feature indicators are acquired, and one or more rules estimated to be necessary to achieve the condition related to the target indicator are generated using at least the values. The rule includes at least one sub-rule related to the feature indicator.

The rule generation method according to this embodiment can be executed by the rule generation device 10 described above. First, when the rule generation device 10 receives an instruction to execute rule generation, in step S110, it acquires time series data of multiple indices (indices 1 to N) and condition information. Then, in step S120, the time series data and condition information are input to the first model 142 of the feature index identification unit 140. Then, information indicating one or more of index 1, index 2, ..., index N, the rate of change of index 1, the rate of change of index 2, ..., and the rate of change of index N is output from the first model 142. The feature index identification unit 140 identifies one or more feature indexes (feature indexes 1 to M) based on the output of the first model 142. Next, in step S130, the value of one or more feature indexes and the condition information are input to the second model 162 of the generation unit 160. One or more rules are output from the second model 162. The generating unit 160 generates an output rule using one or more rules output from the second model 162, and outputs the output rule.

As described above, according to this embodiment, the generation unit 160 acquires values of one or more characteristic indices, and generates one or more rules that are estimated to be necessary to achieve the conditions related to the objective indices using at least the values. Therefore, the conditions necessary to achieve the objective can be clarified.

Second Embodiment
14 is a diagram showing an overview of a determination device 20 according to the second embodiment. The determination device 20 according to this embodiment includes a first acquisition unit 220, a second acquisition unit 240, and a determination unit 260. The first acquisition unit 220 acquires values of one or more indices and condition information indicating a condition for a target index. The second acquisition unit 240 acquires one or more rules according to the condition information. Each of the one or more rules includes one or more sub-rules related to the index. The determination unit 260 determines whether or not the acquired values of the one or more indices satisfy at least any of the one or more sub-rules.

This judgment device 20 makes it possible to clarify the conditions necessary to achieve a goal.

Figure 15 is a diagram illustrating an example of the functional configuration of the determination device 20 according to this embodiment. In the example of this figure, the determination device 20 further includes a rule memory unit 242 and an output unit 280. In this embodiment, one or more rules are stored in advance in the rule memory unit 242. Then, the second acquisition unit 240 reads and acquires one or more rules from the rule memory unit 242. This figure shows an example in which the rule memory unit 242 is included in the determination device 20, but the rule memory unit 242 may be provided outside the determination device 20.

The rules stored in the rule memory unit 242 can be generated using the rule generation device 10 relating to the first embodiment.

The determination device 20 according to this embodiment can be realized using a computer 1000 having a configuration exemplified in FIG. 12, similar to the rule generation device 10 according to the first embodiment. The storage device 1080 stores program modules that realize each functional component of the determination device 20. The processor 1040 reads each of these program modules into the memory 1060 and executes them to realize the function corresponding to each program module. The determination device 20 may be realized by one computer 1000, or may be realized by the cooperation of multiple computers 1000. In addition, when the rule memory unit 242 is provided inside the determination device 20, for example, the rule memory unit 242 can be realized using the storage device 1080.

<First acquisition unit 220>
The first acquisition unit 220 acquires one or more index values, for example, by accepting an input from a user. Alternatively, the first acquisition unit 220 may acquire one or more index values by reading them from a storage device accessible from the first acquisition unit 220 (for example, the storage device 1080), or may acquire them from other external devices. The index values acquired by the first acquisition unit 220 may correspond to, for example, the feature index values acquired by the generation unit 160 in the first embodiment, but are not limited thereto.

The first acquisition unit 220 may acquire one value for each index. Furthermore, the number of indexes for which the first acquisition unit 220 acquires values is not particularly limited, and may be, for example, 100 or less, or 50 or less. The number of indexes for which the first acquisition unit 220 acquires values may be, for example, 10 or more, or 20 or more.

In addition, the first acquisition unit 220 acquires condition information. The condition information is, for example, information indicating the value of the target index or the range of the target index. The target index may be any of the indexes for which the first acquisition unit 220 acquires a value, or may be an index different from the index for which the first acquisition unit 220 acquires a value.

The condition information is input to the determination device 20 by, for example, a user, and the first acquisition unit 220 can acquire the input condition information. For example, options for the target index are displayed on the display of the determination device 20, and the user can input information indicating the target index to the determination device 20 by performing an operation to select one of the options. Also, for example, options indicating the value or range of the target index are displayed on the display of the determination device 20, and the user can input information indicating the value or range of the target index to the determination device 20 by performing an operation to select one of the options.

In addition, the condition information may be stored in advance in a storage device (e.g., storage device 1080) accessible from the first acquisition unit 220, and the first acquisition unit 220 may read and acquire it, or the first acquisition unit 220 may acquire the condition information from an external device.

In addition, for example, the first acquisition unit 220 may first acquire condition information, and then it may be determined for which index a value should be acquired according to the condition information. In this case, for which index a value should be acquired for each objective index is determined in advance. Then, when the first acquisition unit 220 acquires the condition information, it causes an input index corresponding to the objective index indicated in the condition information to be displayed on the display of the determination device 20. Then, the user inputs a value for the displayed index.

<Second acquisition unit 240>
The second acquisition unit 240 acquires one or more rules according to the condition information acquired by the first acquisition unit 220. The one or more rules acquired by the second acquisition unit 240 may be independent of each other as in the example shown in Fig. 7, or may be connected to each other with "AND" or "OR" as shown in Figs. 8 to 11. Hereinafter, the one or more rules acquired by the second acquisition unit 240 are collectively referred to as a rule set. For example, each of Figs. 7 to 11 shows one rule set.

A plurality of rule sets are stored in the rule storage unit 242. Furthermore, rule set reference information is stored in advance in the rule storage unit 242. In the rule set reference information, each rule set is associated with a combination of a target index and a value or range of the target index.

Figure 16 is a diagram illustrating rule set reference information. Using such rule set reference information, the second acquisition unit 240 can identify which rule set should be used based on the condition information. For example, the second acquisition unit 240 identifies a rule set corresponding to the target index indicated in the condition information and the range to which the value of the target index belongs. Then, the second acquisition unit 240 reads and acquires the identified rule set from the rule storage unit 242. In this figure, an example is shown in which the range of the target index is associated with the rule set, but the value of the target index may be associated with the rule set. When the first acquisition unit 220 acquires the condition information via the options as described above, it is preferable that the range or value of the target index in the rule set reference information corresponds to the options.

<Determination unit 260>
The determination unit 260 performs a determination on at least the subrules included in the rule set acquired by the second acquisition unit 240. Specifically, the determination unit 260 determines whether or not the value of the index acquired by the first acquisition unit 220 satisfies each of the subrules acquired by the second acquisition unit 240. Each subrule indicates a condition related to any of the indexes acquired by the first acquisition unit 220. Furthermore, the determination unit 260 may further determine whether or not at least any of the rules are satisfied based on the determination result of the subrules. Specifically, when all subrules included in a certain rule are satisfied, the rule is determined to be satisfied. On the other hand, a rule including even one unsatisfied subrule is determined to be unsatisfied. The determination unit 260 may perform a determination on only some of the subrules included in the rule set, but it is preferable to perform a determination on all the subrules.

In addition, the rule set acquired by the second acquisition unit 240 may include subrules for all indicators for which the first acquisition unit 220 has acquired values, or may include subrules for only some of the indicators.

<Output Unit 280>
The output unit 280 outputs one or more subrules that are determined by the determination unit 260 not to be satisfied by the index value acquired by the first acquisition unit 220. By doing so, the user of the determination device 20 can understand the conditions that need to be satisfied, and can consider countermeasures or take appropriate measures. It is preferable that the output unit 280 outputs all subrules that are determined not to be satisfied by the index value acquired by the first acquisition unit 220. The output unit 280 may further output subrules that are determined to be satisfied by the index value acquired by the first acquisition unit 220. The output method of the output unit 280 is not particularly limited, and may be, for example, display on a display or output to an external device or a storage device.

Figure 17 is a flowchart illustrating the flow of the determination method according to this embodiment. The determination method according to the embodiment is executed by one or more computers. The determination method according to this embodiment includes steps S210, S220, and S230. In step S210, values of one or more indicators and condition information indicating conditions for the target indicator are acquired. In step S220, one or more rules are acquired according to the condition information. Each of the one or more rules includes one or more sub-rules related to the indicator. In step S230, it is determined whether the acquired values of the one or more indicators satisfy at least any of the one or more sub-rules.

The determination method according to this embodiment can be executed by the determination device 20 described above. For example, when a user inputs an index value and condition information to the determination device 20, the first acquisition unit 220 acquires them (step S210). Then, in step S220, the second acquisition unit 240 acquires one or more rules (rule sets) based on the condition information. Then, in step S230, the determination unit 260 determines whether or not a sub-rule included in the rule set is satisfied by the index value acquired by the first acquisition unit 220.

Next, the action and effect of this embodiment will be described. In this embodiment, the same action and effect as in the first embodiment can be obtained. In addition, according to this embodiment, the determination device 20 determines whether the value of one or more acquired indicators satisfies at least one of one or more subrules. Therefore, it is possible to more clearly grasp the indicators and conditions that require measures.

Third Embodiment
18 is a block diagram illustrating a functional configuration of a determination device 20 according to the third embodiment. The determination device 20 according to this embodiment is the same as the determination device 20 according to the second embodiment, except that it further includes a rule generating unit 250 that generates one or more rules. In this embodiment, the second acquisition unit 240 acquires one or more rules generated by the rule generating unit 250. In this embodiment, the determination device 20 does not need to include a rule storage unit 242. This will be described in detail below.

In the example of this figure, the rule generation unit 250 includes a feature index identification unit 251 and a generation unit 252. The feature index identification unit 251 is the same as the feature index identification unit 140 described in the first embodiment, and the generation unit 252 is the same as the generation unit 160 described in the first embodiment. However, the rule generation unit 250 may include the generation unit 252 without including the feature index identification unit 251. The feature index identification unit 251 includes a first model 253, and the generation unit 252 includes a second model 254. The first model 253 is the same as the first model 142 described in the first embodiment, and the second model 254 is the same as the second model 162 described in the first embodiment.

When the rule generating unit 250 does not include the feature index identifying unit 251 but includes the generating unit 252, the rule generating unit 250 generates rules as follows. The generating unit 252 obtains the values of one or more indices obtained by the first obtaining unit 220, and generates one or more rules that are estimated to be necessary to achieve the conditions related to the target index indicated in the condition information, using at least the values. Specifically, the values of one or more indices and the condition information obtained by the first obtaining unit 220 are input to the second model 254 of the generating unit 252. Then, a plurality of rules are output from the second model 254. Then, the generating unit 252 generates output rules in the same manner as the generating unit 160. The second obtaining unit 240 obtains the output rules generated by the generating unit 252.

When the rule generating unit 250 includes a feature index identifying unit 251 and a generating unit 252, the first acquiring unit 220 acquires time series data of multiple indices in the same manner as the acquiring unit 120 of the first embodiment. The feature index identifying unit 251 identifies feature indexes in the same manner as the feature index identifying unit 140 described in the first embodiment. The generating unit 252 acquires values of one or more feature indexes in the same manner as the generating unit 160 described in the first embodiment, and uses at least the values to generate one or more rules that are estimated to be necessary to achieve the conditions related to the target index. In this case, the determination device 20 may include the rule generating device 10 described in the first embodiment.

Next, the action and effect of this embodiment will be described. In this embodiment, the same action and effect as in the first embodiment can be obtained. In addition, the determination device 20 according to this embodiment further includes a rule generation unit 250 that generates one or more rules. Therefore, there is no need to generate and store many rules in advance, and rules can be generated each time according to the condition information.

(Fourth embodiment)
19 is a diagram for explaining the determination device 20 according to the fourth embodiment. This diagram is displayed, for example, on a display of the determination device 20. The determination device 20 according to this embodiment is the same as the determination device 20 according to the second or third embodiment, except for the points described below.

In the determination device 20 according to this embodiment, the output unit 280 outputs one or more unsatisfied subrules. An unsatisfied subrule is a subrule that is determined by the determination unit 260 to not be satisfied by the values of one or more indicators acquired by the first acquisition unit 220. The first acquisition unit 220 then accepts selection of at least one of the one or more output unsatisfied subrules. The first acquisition unit 220 also regards the selected unsatisfied subrule as condition information.

According to such a judgment device 20, for example, when an unsatisfied sub-rule is an indicator or condition that is difficult to directly approach through measures, etc., the user can identify an alternative rule that should be satisfied instead of the unsatisfied sub-rule.

That is, the first acquisition unit 220 performs a new flow from steps S210 to S230 using the unsatisfied subrule as condition information. In this flow, the second acquisition unit 240 acquires an alternative rule to replace the unsatisfied subrule, and a judgment is made on the alternative rule. As a result, the user can grasp the rule that should be satisfied instead of the unsatisfied subrule, and among those rules, the rule that is not satisfied. Such substitution of unsatisfied subrules may be repeated. Therefore, the user can find indicators and conditions that are easier to approach through measures.

This figure illustrates an example of an image displayed on a display as an output by the output unit 280. For example, in the image, a satisfied subrule and an unsatisfied subrule are displayed in different colors. A satisfied subrule is a subrule that is determined by the determination unit 260 when one or more index values acquired by the first acquisition unit 220 are satisfied. In this figure, a satisfied subrule is displayed in white letters on a black background, and an unsatisfied subrule is displayed in black letters on a white background. Next to the unsatisfied subrule, a button 300 ("Generate Alternative Rule" button) for requesting the generation of an alternative rule is displayed. When the user clicks this button 300, the first acquisition unit 220 accepts the selection of the unsatisfied subrule and acquires the selected unsatisfied subrule as condition information.

When the output unit 280 outputs multiple unsatisfied subrules, the user may select only one of them, or may select two or more. When two or more unsatisfied subrules are selected, the judgment device 20 generates an alternative rule for each unsatisfied subrule and performs processing to make a judgment. In generating an alternative rule, an index value that has already been acquired may be used, or one or more index values may be newly acquired by the first acquisition unit 220.

Next, the action and effect of this embodiment will be described. In this embodiment, the same action and effect as in the first embodiment can be obtained. In addition, according to the determination device 20 of this embodiment, the first acquisition unit 220 accepts a selection of at least one of one or more unsatisfied subrules, and sets the selected unsatisfied subrule as condition information. Therefore, for example, when the unsatisfied subrule is an indicator or condition that is difficult to directly approach by measures, etc., the user can understand an alternative rule that should be satisfied instead of the unsatisfied subrule.

The above describes embodiments of the present invention with reference to the drawings, but these are merely examples of the present invention and various configurations other than those described above can also be adopted.

In addition, in the multiple flow charts used in the above explanation, multiple steps (processing) are described in order, but the order of execution of the steps performed in each embodiment is not limited to the order described. In each embodiment, the order of the steps shown in the figures can be changed to the extent that does not cause any problems in terms of content. In addition, each of the above-mentioned embodiments can be combined to the extent that the content is not contradictory.

A part or all of the above-described embodiments can be described as, but are not limited to, the following supplementary notes.
1. An acquisition means for acquiring time series data of a plurality of indices and condition information indicating a condition related to a target index;
a characteristic indicator specifying means for specifying at least one of the plurality of indicators as a characteristic indicator for the target indicator by using time-series data of the plurality of indicators and the condition information;
a generating means for acquiring values of one or more of the characteristic indicators, and generating one or more rules estimated to be necessary for achieving a condition related to the objective indicator, using at least the acquired values;
A rule generating device, wherein the rule includes at least one sub-rule related to the feature indicator.
2. In the rule generating device according to 1.,
The feature index identifying means
extracting one or more of the multiple indices and change rates representing transitions of the multiple indices over time that affect the target index;
A rule generating device that identifies the characteristic index based on at least one of the extracted index and the rate of change.
3. In the rule generating device according to 2.,
The feature index identifying means
A first model for extracting features;
A rule generation device that uses the first model to extract one or more of the multiple indicators and the rate of change that have an effect on the target indicator.
4. In the rule generation device according to any one of 1 to 3,
The generating means is a rule generating device including a second model by ensemble machine learning.
5. In the rule generation device according to any one of 1 to 4,
A rule generation device in which the one or more rules generated include two or more rules that are replaceable with each other.
6. In the rule generation device according to any one of 1 to 5,
The rule generating device, wherein the indicator is an economic indicator.
7. A first acquisition means for acquiring one or more index values and condition information indicating a condition for the target index;
a second acquisition means for acquiring one or more rules, each of which includes one or more subrules related to the index, in response to the condition information;
A determination device comprising: a determination means for determining whether or not the acquired values of the one or more indicators satisfy at least any of the one or more subrules.
8. In the determination device according to 7.,
The one or more rules are stored in advance in a storage means,
The second acquisition means is a determination device that reads and acquires the one or more rules from the storage means.
9. In the determination device according to 7.,
A rule generating means for generating the one or more rules,
The second acquisition means is a determination device that acquires the one or more rules generated by the rule generation means.
10. In the determination device according to any one of 7. to 9.,
The determination device further comprises an output unit that outputs one or more unsatisfied subrules, which are the subrules determined by the determination unit to have one or more index values that are not satisfied.
11. In the determination device according to 10.,
The first acquisition means is
Accepting a selection of at least one of the one or more unsatisfied subrules outputted;
The determination device sets the selected unsatisfied subrule as the condition information.
12. One or more computers:
Acquire time series data of a plurality of indices and condition information indicating a condition related to a target index;
Using the time-series data of the plurality of indices and the condition information, identify at least one of the plurality of indices as a characteristic indices for the target indices;
obtaining values of one or more of the characteristic indicators, and using at least the values to generate one or more rules that are estimated to be necessary to achieve a condition related to the objective indicator;
A rule generating method, wherein the rule includes at least one sub-rule related to the feature indicator.
13. In the rule generation method according to 12.,
the one or more computers:
extracting one or more of the multiple indices and change rates representing transitions of the multiple indices over time that affect the target index;
A rule generation method for identifying the characteristic indicator based on at least one of the extracted indicator and the rate of change.
14. In the rule generation method according to 13.,
A rule generation method in which the one or more computers use a first model for extracting features to extract one or more of the multiple indicators and the rate of change that affect the target indicator.
15. In the rule generation method according to any one of 12. to 14.,
A rule generation method in which the one or more computers generate the one or more rules using a second model by ensemble machine learning.
16. In the rule generation method according to any one of 12. to 15.,
A rule generation method, wherein the one or more rules to be generated include two or more rules that are mutually replaceable.
17. In the rule generation method according to any one of 12. to 16.,
The rule generation method, wherein the indicator is an economic indicator.
18. One or more computers:
Obtaining values of one or more indicators and condition information indicating a condition for the objective indicator;
Obtaining one or more rules, each including one or more subrules related to the indicator, in response to the condition information;
A method for determining whether or not the acquired values of the one or more indicators satisfy at least any of the one or more subrules.
19. In the method according to 18.,
The one or more rules are stored in advance in a storage means,
The one or more computers read and acquire the one or more rules from the storage means.
20. In the determination method according to 18.,
The one or more computers:
generating the one or more rules;
A determination method for obtaining the one or more generated rules.
21. In the method according to any one of 18. to 20.,
The one or more computers output one or more unsatisfied subrules, which are subrules that are determined to not satisfy the values of the one or more indicators.
22. In the determination method according to 21.,
The one or more computers:
Accepting a selection of at least one of the one or more unsatisfied subrules outputted;
A determination method for determining that the selected unsatisfied subrule is the condition information.
23. A computer
An acquisition means for acquiring time series data of a plurality of indices and condition information indicating a condition related to the target index;
a characteristic indicator specifying means for specifying at least one of the plurality of indicators as a characteristic indicator for the target indicator using the time-series data of the plurality of indicators and the condition information; and a generating means for acquiring values of one or more of the characteristic indicators and generating one or more rules estimated to be necessary for achieving a condition related to the target indicator using at least the values.
The program, wherein the rules include at least one sub-rule related to the feature index.
24. In the program according to 23.,
The feature index identifying means
extracting one or more of the multiple indices and change rates representing transitions of the multiple indices over time that affect the target index;
A program for identifying the characteristic index based on at least one of the extracted index and the rate of change.
25. In the program according to 24.,
The feature index identifying means
A first model for extracting features;
A program that uses the first model to extract one or more of the multiple indexes and the rate of change that affect the target index.
26. In the program according to any one of 23 to 25,
The generating means is a program including a second model by ensemble machine learning.
27. In the program according to any one of 23 to 26,
A program in which the one or more rules generated include two or more rules that are mutually replaceable.
28. In the program according to any one of 23 to 27,
The program wherein the indicator is an economic indicator.
29. A computer
A first acquisition means for acquiring one or more index values and condition information indicating a condition for the target index;
a second acquisition means for acquiring one or more rules, each of which includes one or more subrules related to the index, in accordance with the condition information; and a program that functions as a determination means for determining whether the acquired values of the one or more indexes satisfy at least any of the one or more subrules.
30. In the program according to 29.,
The one or more rules are stored in advance in a storage means,
The second acquisition means is a program that reads and acquires the one or more rules from the storage means.
31. In the program according to 29.,
causing the computer to further function as a rule generating means for generating the one or more rules;
The second acquisition means is a program that acquires the one or more rules generated by the rule generation means.
32. In the program according to any one of 29. to 31.,
A program that causes the computer to further function as an output unit that outputs one or more unsatisfied subrules, which are subrules that have been determined by the determination unit to have one or more index values that are not satisfied.
33. In the program according to 32.,
The first acquisition means is
Accepting a selection of at least one of the one or more unsatisfied subrules outputted;
A program for setting the selected unsatisfied subrule as the condition information.

10 Rule generating device 20 Judgment device 120 Acquisition unit 122 Index storage unit 140 Feature index specification unit 142 First model 160 Generation unit 162 Second model 220 First acquisition unit 240 Second acquisition unit 242 Rule storage unit 250 Rule generating unit 251 Feature index specification unit 252 Generation unit 253 First model 254 Second model 260 Judgment unit 280 Output unit 1000 Computer

Claims (8)

An acquisition means for acquiring time series data of a plurality of indices and condition information indicating a condition related to a target index;
a characteristic indicator specifying means for specifying at least one of the plurality of indicators as a characteristic indicator for the target indicator by using time-series data of the plurality of indicators and the condition information;
a generating means for acquiring values of one or more of the characteristic indicators, and generating, using at least the values, one or more rules that are estimated to be necessary for achieving a condition related to the objective indicator;
A rule generating device, wherein the rule includes at least one sub-rule related to the feature indicator. 2. The rule generating device according to claim 1,
The feature index identifying means
extracting one or more of the multiple indices and change rates representing transitions of the multiple indices over time that affect the target index;
A rule generating device that identifies the characteristic index based on at least one of the extracted index and the rate of change. 3. The rule generating device according to claim 2,
The feature index identifying means
A first model for extracting features;
A rule generation device that uses the first model to extract one or more of the multiple indicators and the rate of change that have an effect on the target indicator. 4. The rule generating device according to claim 1,
The generating means is a rule generating device including a second model by ensemble machine learning. 5. The rule generating device according to claim 1,
A rule generation device in which the one or more rules generated include two or more rules that are replaceable with each other. 6. The rule generating device according to claim 1,
The rule generating device, wherein the indicator is an economic indicator. One or more computers
Acquire time series data of a plurality of indices and condition information indicating a condition related to a target index;
Using the time-series data of the plurality of indices and the condition information, at least one of the plurality of indices is identified as a characteristic indices for the target indices;
obtaining values of one or more of the characteristic indicators, and using at least the values to generate one or more rules that are estimated to be necessary to achieve a condition related to the objective indicator;
A rule generating method, wherein the rule includes at least one sub-rule related to the feature indicator. Computer,
An acquisition means for acquiring time series data of a plurality of indices and condition information indicating a condition related to the target index;
a characteristic indicator specifying means for specifying at least one of the plurality of indicators as a characteristic indicator for the target indicator using the time-series data of the plurality of indicators and the condition information; and a generating means for acquiring values of one or more of the characteristic indicators and generating one or more rules estimated to be necessary for achieving a condition related to the target indicator using at least the values.
The program, wherein the rules include at least one sub-rule related to the feature index.

Images