Introduction to widely used regression models in medical research using R, STATA, and SPSS : A tutorial

Abstract

Abstract Background: Regression models are often used to discover the relationship between variables. According to the type and scale of response variables, there are different type of regression models. Determining the type of regression is difficult for medical researchers. Methods: In this tutorial, How to use different regressions models (ie, linear Regression, Ridge Regression, Polynomial Regression, Lasso Regression, Bayesian Linear Regression, Principal Components Regression, Partial Least Squares Regression, Elastic Net Regression, Support Vector Regression, Logistic Regression, Quantile Regression, Ordinal Regression, Poisson regression, Negative binomial regression, Quasi Poisson Regression, Fractional Regression, Cox Regression, Tobit Regression ), their computational difficulties and their assumptions are shown. The application of each regression model is also specified with a medical example, then the model is implemented in different software (ie, SPSS, STATA, R) and its output is described. Results: The models in this study are introduced according to the dependent variable’s type and scale; some models were used to increase the efficiency and improve the estimation of the relationship between specific types of variables. Conclusion: Today, the use of regression models is growing and new statistical methods such as data mining basically use regression as a tool for forecasting. In this tutorial, we tried to help researchers to gain a better understanding of the computational implementation of different regression models by introducing widely used regression models, providing software instructions and interpretation.