A method for predict user churn unbalance data for mobile communication

A method for predict loss unbalance data of mobile communication users is provided. This method is mainly based on the improved depth forest model algorithm framework to classify the unbalanced data of communication user churn. First, new parameters are constructed in the multi-granularity window sliding process of the deep forest model to control the sliding of different types of user data. The sliding data is brought into the cascade forest part of the training, and each decision tree in the forest gives different weights to different types of data according to the training results. The final voting result of the algorithm model adopts weighted post-voting, so as to realize the processing of unbalanced user data. The prediction method provided by the invention improves the multi-granularity sliding module in the depth forest to slide different classifications, and at the same time, each decision tree in the forest updates the weights of different types of users, so that the whole model has higher identifica