Machine Learning – Reviews system Source – datasciencecentral.com
I am working on the attached data set where I have to predict the reviews (1-5). Date is the only column to play with and I have extracted day/month and year out of it. After fitting random forest, I am getting .85 accuracy on training and .80 on test. Clear case of over fitting. If any one provide me with some insights on how to improve the accuracy or with some feature engineering would be great.