A Method for Predicting the Macro Environment of Enterprise Management Based on Linear Models

Abstract

Aiming at interval data, this paper combines the additive model on the basis of the partial linear model and introduces the sliding window model, and proposes an additive linear model based on the midpoint and range of each window of the sliding window mode. This model combines the advantages of semi-parametric regression and sliding window model, and avoids the dimensional disaster. In addition, the model determines the number of phases of the sliding window according to the cross-entropy criterion, and gives an iterative algorithm for estimating model parameters and unknown function based on the least square method and kernel estimation method. In the empirical analysis, several financial indicators are introduced to predict the law of macroeconomic development. The results show that the improved model is better than the traditional regression model.