DEVICE AND METHOD FOR CONSTRUCTING DATA MINING MODEL USING ENSEMBLE MODEL

摘要

PURPOSE: A device and a method for constructing a data mining model using an ensemble model are provided to calculate the analysis and the prediction of the multidimensional huge data faster than a current method, and to construct a stable model easily. CONSTITUTION: The learning data, which is the multidimensional data, is inputted. A weight is calculated by using a residual of the ensemble model. A probability that a reaction variable for a description variable x given from the inputted data gets to be 1 is predicted by using a basic learning device and the weight(S1303). A modification seed tree making a loss function given to the predicted probability minimal is calculated(1304). The learning device is updated and reconstructed by using the calculated modification seed tree(S1305).