The aim of this report is to address the problem of predicting a numerical target variable, starting from categorical, numerical and textual features. The followed approach is based on gradient boosting and sentence vectorization, which perform significantly better than the decision tree regression model used as baseline.
I suggest you to read my report for more info!