PENDUGAAN PARAMETER REGRESI LOGISTIK DENGAN JARINGAN SYARAF TIRUAN

Abstract

Logistic regression can be mapped equivalent to artificial neural network (ANN) without hidden layer with logistic as activation function, hence logistic regression is subset of ANN. The result study on binary and poly-chotomous response data show that parameter estimation values of ANN and logistic regression are similar. In comparison with ANN, logistic regression has standard procedure for estimation and testing parameter.
Keyword : logistic regression, generalized linear model, artificial neural network, activation function, hidden layer