A comparison between the logistic regression model and Linear Discriminant analysis using Principal Component unemployment data for the province of Baghdad

Abstract

The objective of the study is to demonstrate the predictive ability is better between the logistic regression model and Linear Discriminant function using the original data first and then the Home vehicles to reduce the dimensions of the variables for data and socio-economic survey of the family to the province of Baghdad in 2012 and included a sample of 615 observation with 13 variable, 12 of them is an explanatory variable and the depended variable is number of workers and the unemployed. Was conducted to compare the two methods above and it became clear by comparing the logistic regression model best of a Linear Discriminant function written using the original data, either using Principal Component was reduced variables to 5 key factors by 62.875% of the total variance and the results were equal . That the performance of a logistic regression equal to using the original data and Principal Component, while performing a Linear Discriminant function using Principal Component was better than the original data.