Prediction model establishing device and method as well as computer readable storage medium

Abstract

The invention discloses a prediction model establishing device. The prediction model establishing device comprises a storage, a processor and a prediction model establishing program which is stored onthe storage and can operate on the processor, and the program is implemented by the processor to realize the following steps: obtaining exogenous characteristics in one or more time units, before a target time unit, of a to-be-tested item as well as self-return time characteristics before the target time unit; performing pre-processing and normalization processing on the characteristics to obtaina normalized characteristic set; performing characteristic screening on the characteristics according to a preset rule to obtain prediction characteristics; taking a practical observed value of the to-be-tested item as a prediction target, and taking prediction characteristics and the prediction target as a prediction sample; obtaining a plurality of prediction samples through the step, and inputting the prediction samples into a prediction regression model for training to generate the prediction model. The invention further discloses a prediction model establishing method and a computer readable storage medium. The prediction accuracy rate of the prediction module is increased.

Description

technical field [0001] The present invention relates to the technical field of terminals, and in particular to a predictive model building device, method and computer-readable storage medium. Background technique [0002] At present, the technology of predicting data through machine learning has been applied in more and more fields, such as the prediction of advertising click-through rate, the prediction of the incidence of certain epidemic diseases, etc. The historical data of the project to be predicted constitutes a time series, and an autocorrelation regression time series model (ARIMA) is established based on the characteristics of the time series itself for prediction, but this model only uses the trend characteristics of the project to be predicted to make predictions, and cannot be combined with forecasting external Feature usage, leading to poor prediction accuracy. Contents of the invention [0003] The present invention provides a prediction model establishment...

AI Technical Summary

Problems solved by technology

[0002] At present, the technology of predicting data through machine learning has been applied in more and more fields, such as the prediction of advertising click-through rate, the prediction of the incidence of certain epidemic diseases, etc. The historical data of the project to be predicted constitutes a time series, and an autocorrelation regression time series model (ARIMA) is established based on the characteristics of the time series itself for prediction, but this model only uses the trend characteristics of the project to be predicted to make predictions, and cannot be combined with forecasting external Feature usage, resulting in poor prediction accuracy

Images