ValueError: Input contains NaN, infinity or a value too large for dtype('float32'). randomforest run

Question

#fill -999 to NAs
X = X_train.fillna(-999)
y = y_train.fillna(-999) 

import lightgbm as lgb
import xgboost as xgb

NFOLDS = 8
folds = KFold(n_splits=NFOLDS)

#====================================

xgb_submission=sample_submission.copy()
xgb_submission['isFraud'] = 0
import xgboost as xgb
from sklearn.metrics import roc_auc_score
for fold_n, (train_index, valid_index) in enumerate(folds.split(X)):

    X_train_, X_valid = X.iloc[train_index], X.iloc[valid_index]
    y_train_, y_valid = y.iloc[train_index], y.iloc[valid_index]
    #xgbclf.fit(X_train_,y_train_)

    rf_clf1 = RandomForestClassifier(n_estimators=300, max_depth = 10, min_samples_leaf=8, \
                                    min_samples_split=8, random_state=0)
    rf_clf1.fit(X_train,y_train_)
    pred = rf_clf1.predict(X_test)
    print(pred)

I checked the X or y has any Nan but no
but it gives the error with ValueError: Input contains NaN, infinity or a value too large for dtype('float32').

> print(type(X),type(y)) 
> <class 'pandas.core.frame.DataFrame'> <class'pandas.core.series.Series'>

Answer 1

When does this error appear actually - while assigning X_train_, X_valid values or while fitting the datasets to RandomForest algorithm?

I also see from the code that in the first turn you define X_train_ dataframe:

**X_train_**, X_valid = X.iloc[train_index], X.iloc[valid_index]

Whereas you fit the rf_clf1 object to another dataset (namely: X_train )

rf_clf1.fit(X_train,y_train_)

So here the missing _ in the variable name might be the case as well.