PENDETEKSIAN OUTLIER PADA REGRESI LOGISTIK DENGAN MENGGUNAKAN TEKNIK TRIMMED MEANS

This final project discusses the outlier of logistic regression. The estimator is obtained through maximum likelihood method. Then numerical approach of which is named Newton-Raphson method is applied. Furthurmore goodness of fitting by using Hosmer and Lemeshow test and the coefficient of determination R2 is evaluated to be interprete the predictor variables which are explained by the respond variables. The next step is outlier detection by trimming of outlier estimate data in the side X, this technique is an idea from trimmed means. Trimming of data affects the regression model and upgrade the coefficient of determination R2. The compute and data analysis by using R program version 3.2.5 Keywords: Logistic regression model, outlier, trimmed means, maximum likelihood method, Newton-Raphson method, Hosmer and Lemeshow test, coefficient of determination R2