I'm trying to train the model using Xgboost. The code is doing split using KFold. And for each fold, it's running the Xgboost model using fit. Within the fit function, I'm trying to evaluate both train and valid data to check if the errors. And then doing the prediction in test set.
I'm running the following code using Xgboost.
kf = GroupKFold(n_splits=4)
for trn_idx, test_idx in kf.split(X, groups=X.year) :
x_train, x_valid = X.iloc[trn_idx], X.iloc[test_idx]
y_train, y_valid = y.iloc[trn_idx], y.iloc[test_idx]
xgb_model = xgb.XGBRegressor(
booster = 'dart',
eta = 0.1,
gamma = 0,
colsample_bytree = 0.7,
n_estimators = 1200,
max_depth = 1,
reg_alpha = 1.1,
reg_lambda = 1.1,
subsample = 0.03,
eval_metric=my_smape)
xgb_model.fit(x_train, y_train,
eval_set=[(x_train, y_train), (x_valid,y_valid)], early_stopping_rounds=20,
verbose=True)
But I'm getting the following error. I checked this doc, and my code is according to the doc. Can someone please help me find the solution?
AttributeError Traceback (most recent call last)
<ipython-input-38-81b11a21472c> in <module>
23 eval_metric=my_smape)
24
---> 25 xgb_model.fit(x_train, y_train,
26 eval_set=[(x_valid,y_valid)], early_stopping_rounds=20,
27 verbose=True)
D:\Anaconda\lib\site-packages\xgboost\core.py in inner_f(*args, **kwargs)
573 for k, arg in zip(sig.parameters, args):
574 kwargs[k] = arg
--> 575 return f(**kwargs)
576
577 return inner_f
D:\Anaconda\lib\site-packages\xgboost\sklearn.py in fit(self, X, y, sample_weight, base_margin, eval_set, eval_metric, early_stopping_rounds, verbose, xgb_model, sample_weight_eval_set, base_margin_eval_set, feature_weights, callbacks)
959 xgb_model, eval_metric, params, early_stopping_rounds, callbacks
960 )
--> 961 self._Booster = train(
962 params,
963 train_dmatrix,
D:\Anaconda\lib\site-packages\xgboost\core.py in inner_f(*args, **kwargs)
573 for k, arg in zip(sig.parameters, args):
574 kwargs[k] = arg
--> 575 return f(**kwargs)
576
577 return inner_f
D:\Anaconda\lib\site-packages\xgboost\training.py in train(params, dtrain, num_boost_round, evals, obj, feval, maximize, early_stopping_rounds, evals_result, verbose_eval, xgb_model, callbacks, custom_metric)
180 break
181 bst.update(dtrain, i, obj)
--> 182 if cb_container.after_iteration(bst, i, dtrain, evals):
183 break
184
D:\Anaconda\lib\site-packages\xgboost\callback.py in after_iteration(self, model, epoch, dtrain, evals)
237 for _, name in evals:
238 assert name.find('-') == -1, 'Dataset name should not contain `-`'
--> 239 score: str = model.eval_set(evals, epoch, self.metric, self._output_margin)
240 splited = score.split()[1:] # into datasets
241 # split up `test-error:0.1234`
D:\Anaconda\lib\site-packages\xgboost\core.py in eval_set(self, evals, iteration, feval, output_margin)
1860 if feval is not None:
1861 for dmat, evname in evals:
-> 1862 feval_ret = feval(
1863 self.predict(dmat, training=False, output_margin=output_margin), dmat
1864 )
D:\Anaconda\lib\site-packages\xgboost\sklearn.py in inner(y_score, dmatrix)
99 def inner(y_score: np.ndarray, dmatrix: DMatrix) -> Tuple[str, float]:
100 y_true = dmatrix.get_label()
--> 101 return func.__name__, func(y_true, y_score)
102 return inner
103
AttributeError: '_PredictScorer' object has no attribute '__name__'
It looks like you've run make_scorer() on your custom metric. Try supplying the original function as eval_metric instead, this should fix the issue.