Training method and device of cycle prediction model, cycle prediction method and device

Abstract

The invention relates to a training method and device of a cycle prediction model, a cycle prediction method and device, and relates to the technical field of machine learning. The method comprises the steps of calculating a historical order behavior sequence of a historical user and an actual re-purchase cycle of the historical articles included in a historical order according to the historical user data; constructing a data set according to the historical articles, the actual re-purchase cycle, the historical order behavior sequence and the historical user portraits in the historical user data; inputting the historical articles, the historical order behavior sequence and the historical user portrait included in the data set into a to-be-trained period prediction model to obtain a predicted re-purchase period of the historical article; and constructing a first loss function according to the actual re-purchase cycle and the predicted re-purchase cycle in the data set, and training the to-be-trained cycle prediction model by using the first loss function to obtain a trained cycle prediction model. The method improves the accuracy of the cycle prediction model.

Description

Technical field [0001] Example disclosure relates to a machine learning present technical field, particularly, to a training cycle model prediction method, the prediction model training apparatus cycle period prediction method, a computer-readable storage medium and an electronic device. Background technique [0002] In the conventional cycle prediction model that may be implemented in DNN fully connected neural network structure (DeepNeural Networks, Neural Networks depth) having a three-layer model. Specifically, during training can be user information feature and item information characteristics (e.g., article identification, item price, number of items) corresponding to an input DNN model, calculated to the neural network to obtain user multiplexing available period of the article. [0003] Meanwhile, when the DNN model training, according to the user - the output of the complex product commercially available cycle model and reuse cycle true value constructor function loss, a...

AI Technical Summary

Problems solved by technology

[0007] The purpose of the present disclosure is to provide a training method for a cycle prediction model, a training device for a cycle prediction model, a cycle prediction method, a computer-readable storage medium, and an electronic device, so as to overcome the limitations and defects of related technologies at least to a certain extent. The problem of low accuracy of the cycle forecast model caused by

Images