Detection of Multicollinearity Using Min-Max and Point-Coordinates Approach

Multicollinearity is one of the problems or challenges of modeling or multiple regression usually encountered by Economists and Statisticians. It is a situation where by some of the independent variables in the formulated model are significantly or highly related/correlated. In the past, methods such as Variance Inflation Factor, Eigenvalue and Product moment correlation have been used by researchers to detect multicollinearity in models such as financial models, fluctuation of market price model, determination of factors responsible for survival of man and market model, etc. The shortfalls of these methods include rigorous computation which discourages researchers from testing for multicollinearity, even when necessary. This paper presents moderate and easy algorithm of the detection of multicollinearity among variables no matter their numbers. Using Min-Max approach with the principle of parallelism of coordinates, we are able to present an algorithm for the detection of multicollinearity with appropriate illustrative examples.