I made a comparison between the Linear Regression model and Decision Tree on the problem of predicting house Sale Prices (dataset is also available in the Lin. Reg. example on DataQuest). What I found is that Decision Tree seems to predict the price better especially for high prices where Lin. Reg. seems to underestimate the price. I guess the problem with the data is that most features do not show linear correlation with the price.
I was wondering about your experience with these two models. Any thoughts on this ?