Information processing system, feature description method and feature description program

1. An information processing system comprising: a hardware including a processor; a descriptor generation unit, implemented by the processor, which generates a feature descriptor by using a first template prepared in advance, the feature descriptor generating a feature that affects a prediction target from a first table including a variable of the prediction target and a second table; a feature generation unit, implemented by the processor, which generates the feature by applying the feature descriptor to the first table and the second table; a feature explanation generation unit, implemented by the processor, which generates a feature explanation about the feature descriptor or the feature on the basis of a second template; and an accepting unit, implemented by the processor, which accepts values to be assigned to the first template and the second template; wherein the descriptor generation unit generates the feature descriptor by assigning the accepted values to the first template, the feature explanation generation unit generates the feature explanation by assigning the values assigned to the first template to the second template, and the feature generation unit generates the feature for use as a candidate for an explanatory variable when learning a model for predicting a value of the prediction target.

2. The information processing system according to claim 1 , wherein

the first template and the second template each include

a joint parameter to which a value indicating key columns when joining the first table and the second table is assigned, and

an aggregation parameter to which a value indicating an aggregation operation to be performed on a plurality of rows in the second table and a column as a target of the aggregation operation is assigned, and

the accepting unit accepts joint information indicating a value to be assigned to the joint parameter and aggregation information indicating a value to be assigned to the aggregation parameter.

3. The information processing system according to claim 2 , wherein

the first template and the second template each further include a selection parameter to which a value indicating a condition for selecting any row fulfilling the condition from the second table and a column as a target of the selection is assigned, and

the accepting unit further accepts selection information indicating a value to be assigned to the selection parameter.

4. The information processing system according to claim 2 , wherein

the accepting unit accepts one or more pieces of the joint information and one or more pieces of the aggregation information,

the descriptor generation unit generates a plurality of combinations of the accepted pieces of the joint information and the aggregation information, and generates the feature descriptors for the respective combinations, and

the feature explanation generation unit generates the feature explanations for the respective combinations,

the system further comprising:

a feature explanation appointing unit, implemented by the processor, which appoints the feature explanation to the feature descriptor that has been generated on the basis of the combination used when generating the feature explanation, or appoints the feature explanation to the feature that has been generated by applying the feature descriptor.

5. The information processing system according to claim 4 , wherein

the first template and the second template each further include a selection parameter to which a value indicating a condition for selecting any row fulfilling the condition from the second table and a column as a target of the selection is assigned,

the accepting unit further accepts one or more pieces of selection information indicating a value to be assigned to the selection parameter,

the descriptor generation unit generates a plurality of combinations of the accepted pieces of the joint information, the aggregation information, and the selection information, and generates the feature descriptors for the respective combinations, and

the feature explanation generation unit generates the feature explanations for the respective combinations.

6. The information processing system according to claim 2 , wherein

the first template and the second template each further include a table parameter to which a value identifying the second table is assigned, and

the accepting unit further accepts table information indicating a value to be assigned to the table parameter.

7. The information processing system according to claim 6 , wherein

the second template includes

a table template including a table parameter,

a joint template including a joint parameter,

an aggregation template including an aggregation parameter,

a selection template including a selection parameter, and

a label template including a plurality of label parameters to which labels are assigned, and

the feature explanation generation unit

generates a table label, expressed in natural language, from the table template by assigning table information to the table parameter,

generates a joint label, expressed in natural language, from the joint template by assigning joint information to the joint parameter,

generates an aggregation label, expressed in natural language, from the aggregation template by assigning aggregation information to the aggregation parameter,

generates a selection label, expressed in natural language, from the selection template by assigning selection information to the selection parameter, and

generates the feature explanation from the label template by applying the table label, the joint label, the aggregation label, and the selection label to positions of the respective label parameters.

8. The information processing system according to claim 2 , wherein in a case where rows in the first table and rows in the second table have a one-to-many correspondence when joining the tables, the aggregation parameter is assigned information indicating the aggregation operation to be performed on the plurality of rows in the second table and the column as a target of the aggregation operation.

9. The information processing system according claim 1 , further comprising:

a learning unit, implemented by the processor, which learns a model for predicting a value of the prediction target, by using the generated feature as a candidate for an explanatory variable.

10. The information processing system according to claim 9 , further comprising:

a prediction unit, implemented by the processor, which obtains a predicted value by applying the generated feature to the model.

11. A feature explanation generation method comprising:

generating a feature descriptor by assigning accepted values to a first template prepared in advance, the feature descriptor generating a feature that affects a prediction target from a first table including a variable of the prediction target and a second table;

generating the feature by applying the feature descriptor to the first table and the second table, the feature being used as a candidate for an explanatory variable when learning a model for predicting a value of the prediction target; and

generating a feature explanation about the feature descriptor or the feature by assigning the values assigned to the first template to a second template.