How to Save Gradient Boosting Regressor Results to File?-CodePudding

I have a Gradient Boosting Regressor model for which I would like to save the results to a csv.

from sklearn import ensemble
clf = ensemble.GradientBoostingRegressor(n_estimators = 400, max_depth = 5, min_samples_split = 2,
          learning_rate = 0.1, loss = 'squared_error')
clf.fit(x_train, y_train)

I would like to see the actuals and predicted values.

CodePudding user response：

After fitting your model, you need to actually predict the predictions, with clf.predict, lets say that we want to save y_pred_train & y_pred_test:

y_pred_train = clf.predict(y_train)
y_pred_test = clf.predict(y_test)

Now that we have both predicted values and the actual ones for both train & test sets, we can save a .csv file with np.savetxt:

np.savetxt("train_preds.csv", np.column_stack((y_train, y_pred_train)), delimiter=",", fmt='%.5f')

And

np.savetxt("test_preds.csv", np.column_stack((y_test, y_pred_test)), delimiter=",", fmt='%.5f')

We are using np.column_stack just to stack them as columns rather as rows since they are 1D numpy arrays in their nature.

Please Note that I formatted the floating point precision just for visibility purposes, if you are going to calculate the loss afterwards it's better not to format at all.