Neural Network Modeling and Multiple Linear Regression Modeling in Data Trend Prediction

Abstract

This paper is based on neural network model and multiple linear regression model around data trend prediction. Firstly, feature engineering is carried out to provide rich data dimensions for the follow-up by extracting various features such as base features, contribution rate features, athlete features, etc. Second, the feed-forward neural network-based regression model was selected for medal data prediction after testing various models, and entropy weighting and hierarchical analysis were also used to construct models for specific data prediction; at the same time, multivariate linear regression models were used to evaluate the performance of the models in conjunction with K-fold cross validation. Finally, these models not only have high goodness-of-fit in data prediction, but also have high accuracy and credibility in prediction results, which is a significant advantage in data prediction.