A comparative analysis of neural network and decision trees in forecasting

In a wide range of analysis, decision trees and neural networks are used for predictive analysis. There are researchers who compared the performance of decision trees and Neural networks, however very few with Classification and Regression Tree (CART) and Neural networks (NNs). Here we performed a three way comparison of CART, Random Forest (RF) and NNs models using a continuous and categorical dependent variable for prediction. A web usage sales data set is used to run these models. Measurement of different predictive accuracy methods are used to compare the performance of the models. Experimental results of test data of the model is used here to predict the accuracy.