mVIF Package: A Tool for Detecting Multicollinearity without Dependent Variables

Abstract

— This article discusses the Variance Inflation Factor (VIF), a tool used to test the assumption of non-multicollinearity in regression analysis. VIF measures the correlation between variables in a regression model and its impact on the accuracy of analysis results. The article highlights that VIF can also be used to determine the presence of multicollinearity among variables in various types of analyses, including Hierarchical Cluster Analysis. While there are several programs or packages available to calculate VIF, they usually require a dependent variable input. To address this issue, the author aims to create a new package using Python to calculate VIF without the need for a dependent variable input. The program calculates VIF using the sequential elimination method, which involves removing one variable at each iteration of the for loop. In use, the user needs to input data in the form of a matrix, and the program will return a list of VIFs and information about the presence of multicollinearity in the data. The program provides an alternative method for evaluating multivariate data and the presence of multicollinearity, making the testing process easier and faster for data analysts and researchers.