Prediction interpretation apparatus and prediction interpretation method
A prediction interpretation apparatus, comprising: a data storage unit configured to store data of a plurality of users; a model storage unit configured to store a prediction model learned from data of the whole of the plurality of users; a vicinity user search unit configured to extract vicinity users for the target user from the data storage unit; a linear regression model learning unit configured to learn a linear regression model approximated to the prediction model for the vicinity users; and an interpretation result output unit configured to output an interpretation result of prediction for the target user based on a partial regression coefficient of the linear regression model, wherein the vicinity user search unit extracts the vicinity user by narrowing vicinity user candidates extracted based on distance between users based on a prediction direction of the target user by the prediction model.
1 . A prediction interpretation apparatus, comprising:
a memory configured to store data of a plurality of users,
wherein the memory is configured to store a prediction model learned from data of the plurality of users; and
a processor configured to extract final vicinity users for a target user from the memory,
wherein the processor is configured to learn a linear regression model approximated to the prediction model for the final vicinity users;
wherein the processor is configured to output interpretation results of prediction for the target user based on one or more partial regression coefficients of the linear regression model;
wherein the interpretation results comprise one or more life habits presented to the target user in descending order based upon an impact between a life habit of the one or more life habits and one or more health risks of the target user;
wherein the processor extracts the final vicinity users by narrowing vicinity user candidates extracted based on a distance between users based on a prediction direction of the target user by the prediction model, and
wherein the processor is configured to narrow the vicinity user candidates by excluding at least one vicinity user candidate from the final vicinity users based upon a deviation between a prediction value associated with the at least one vicinity user candidate and the prediction direction of the target user.
2 . The prediction interpretation apparatus as claimed in claim 1 , wherein the processor:
calculates a differential coefficient of the prediction model as the prediction direction,
calculates a slope of a line connecting a point corresponding to a state for one of the vicinity user candidates and a prediction result according to the prediction model, and a point corresponding to a state for the target user and a prediction result according to the prediction model, and
extracts a vicinity user candidate whose product of the differential coefficient and the slope of the line is not negative as a vicinity user.
3 . The prediction interpretation apparatus as claimed in claim 2 ,
wherein the processor outputs one or more items corresponding to each of the one or more partial regression coefficients based on an order of magnitude of the one or more partial regression coefficients of the linear regression model.
4 . The prediction interpretation apparatus as claimed in claim 1 ,
wherein the processor outputs one or more items corresponding to each of the one or more partial regression coefficients based on an order of magnitude of the one or more partial regression coefficients of the linear regression model.
5 . The prediction interpretation apparatus as claimed in claim 4 ,
wherein, when a value of a certain item is changed, the processor outputs a changed prediction result based on the one or more partial regression coefficients.
6 . A prediction interpretation method executed by a prediction interpretation apparatus including a memory configured to store data of a plurality of users; wherein the memory is configured to store a prediction model learned from data of the plurality of users, the prediction interpretation method comprising:
extracting final vicinity users for a target user from the memory;
learning a linear regression model approximated to the prediction model for the final vicinity users; and
outputting interpretation results of prediction for the target user based on one or more partial regression coefficients of the linear regression model,
wherein the interpretation results comprise one or more life habits presented to the target user in descending order based upon an impact between a life habit of the one or more life habits and one or more health risks of the target user, and
wherein the prediction interpretation apparatus extracts the final vicinity users by narrowing vicinity user candidates extracted based on a distance between users based on a prediction direction of the target user by the prediction model, and
wherein narrowing the vicinity user candidates comprises excluding at least one vicinity user candidate from the final vicinity users based upon a deviation between a prediction value associated with the at least one vicinity user candidate and the prediction direction of the target user.