Partial Least Square (PLS) Method of Addressing Multicollinearity Problems in Multiple Linear Regressions (Case Studies: Cost of electricity bills and factors affecting it)
D W Wondola, Salmon N. Aulele, Ferry Kondo Lembang
Abstract
Abstract Multiple regression analysis is a statistical analysis used to predict the effect of several independent variables on the dependent variable. The problem that often occurs in multiple linear regression models is multicollinearity which is a condition of a strong relationship between independent variables. To overcome the problem of multicollinearity, the Partial Least Square method is used. This method reduces independent variables that have no significant effect on the dependent variable, then new variables with smaller dimensions are formed which are linear combinations of the independent variables, therefore the partial significance test (t test) becomes an important part in the formation of PLS components. Furthermore, using the PLS method, we obtain: Ŷ = 126.220 + 12.034 (Income) + 12.437 (Number of Family Members) + 12.959 (House Area) +11.919 (Number of Rooms) +12.274 (Number of Electronic Devices)