Trying Random Forest from Machine Learning course on Kaggle Restaurant Revenue Prediction


I am currently going through Machine Learning Videos. From Lesson 1 which introduces Random Forest i am trying Restaurant Revenue Prediction from old Kaggle competition which is more or less same as Blue Book for Bulldozers example which is done in lesson1.
Objective of this competition was to predict the annual restaurant sales. Using the features set like Restaurant id, date opened, location, city type, restaurant type, coded category from P1-P37 and the dependent variable Revenue.
I pretty much tried to follow the same steps as used in Lesson1 however validation and test scores are very poor even after trying different numbers for n_estimators and max_features.

Kindly let me know what i could try differently in Random Forest to get better results.

TFI_lesson1.slides.pdf (147.4 KB)

Thank You.