Predict when a customer would buy from previous buying history (Mining Buying patterns)

Question

I'm trying to predict when a customer would buy a product(which month of the year) based on the historical transactions. I have 3 years transactions data and good amount of customers are frequent ones. Or basically want to mine the historical transaction data, and see for a set of customers (in a Geo/type of customer etc.) which months are likely period of buying a product X.

Can someone please advice which Machine learning technique or statistical approach would be best in this case.

I am trying - Decision tree classification/Logistic regression model taking month as predictor variable. - Statistical analysis to see which month is more significant for a set of customers.

Answer 1

I would try using xgboost classifier. In my case, xgboost outperformed Random Forest.

import xgboost as xgb
xgb_model = xgb.XGBClassifier(
    learning_rate=0.1, n_estimators=1000, max_depth=3, min_child_weight=1,
    gamma=0, subsample=0.8, colsample_bytree=0.8, objective='binary:logistic',
    nthread=4, scale_pos_weight=1, seed=27)

You can use the below snippet to avoid over fitting problem.

xgb_model = xgb_model.fit(X_train, y_train.astype(int), early_stopping_rounds=50,
                          eval_set=[(X_test, y_test['target'])])