Review of Regression Models and Their Characteristics

Abstract

he paper provides an overview of regression models commonly used in statistics and

machine learning, emphasizing their importance in predicting and understanding

relationships between variables across diverse datasets. The models covered include

Linear Regression (LR), polynomial regression (PR), Decision Tree Regression (DTR),

Neural Network Regression (NNR), Random Forest Regression, and Support Vector

Regression (SVR), With the introduction of special models of regression models, they

are Time Series Regression (TSR) and Spatial Linear Regression (SLR). Linear

regression focuses on linear relationships, while polynomial regression captures

nonlinear patterns by introducing polynomial terms. Decision trees and random forests,

as ensemble methods, partition data recursively, while support vector regression uses

support vector coefficients and kernel functions to handle nonlinear relationships.

GRNN is a fast and efficient model in certain situations and may struggle with

performance on large datasets, while it is more flexible and easy to customize, but

requires intensive training to achieve outstanding results. Time series regression (TSR)

is a powerful tool for modeling and forecasting time-dependent data while spatial linear

regression (SLR) is a powerful extension of traditional linear regression that

incorporates spatial relationships, enabling it to analyze and forecast spatial data.