I'm doing regression analysis on predicting number of products to order during promotion period. However, my numerical features only consist of discount amount and temperature while I have so many categorical data (such as one hot encoding marking the DOW, months, holiday etc).
To be more specific, all my categorical data are just binary data. There are 30 categorical features and 2 numerical features. I have around 900 sample size (approx 2 years and a half)
I've tried using linear regression/ neural network / random forest regression and I've found that neural network is very sensitive to the change of 0,1 in categorical feature)
I was wondering if doing regression model with small number of numerical features and large number of categorical features would be a problem. And if so, could you please suggest the solution with this issue? Thank you very much
