CODEWITHSUNDEEP
pythonmachine-learningscikit-learnxgboostgrid-search
Alex T
Alex T

Reputation: 3754

Plot MAE, RMSE in XGboost model

I'm trying to plot MAE and RMSE from the XGboost model results. First I used gridsearchcv to find params then I fit the model and set eval_metrics to be printed out when fitting the model:

myModel = GridSearchCV(estimator=XGBRegressor(
                        learning_rate=0.01,
                        n_estimators=500,
                        max_depth=5,
                        min_child_weight=5,
                        gamma=0,
                        subsample=0.8,
                        colsample_bytree=0.8, 
                        eval_metric ='mae',
                        reg_alpha=0.05
                        ),
                       param_grid = param_search,
                       cv = TimeSeriesSplit(n_splits=5),n_jobs=-1
                      )

#Fit model
eval_set = [(X_train, y_train), (X_test, y_test)]
eval_metric = ["rmse","mae"]
history=myModel.fit(X_train, y_train, eval_metric=eval_metric, eval_set=eval_set)

I get correct result of this fit:

[0] validation_0-rmse:7891  validation_0-mae:7791.42    validation_1-rmse:6465.99   validation_1-mae:6465.52
[1] validation_0-rmse:7813.98   validation_0-mae:7714.55    validation_1-rmse:6398.87   validation_1-mae:6398.4

However I tried accessing those values in order to create a plot but I get the following error:

myModel.evals_result()

AttributeError: 'GridSearchCV' object has no attribute 'evals_result'

How can I access those values?

Upvotes: 1

Views: 3979

Answers (1)

joyzaza
joyzaza

Reputation: 186

You can create a result dict then pass it to fit

progress = dict()

history=myModel.fit(X_train, y_train, evals_result=progress eval_metric=eval_metric, eval_set=eval_set)

print(progress)

Upvotes: 2

Related Questions