How to solve XGBoost SoftmaxMultiClassObj: label must be in [0, num_class) error?

Question

I'm working on building a model using XGBoost to predict corona virus infections based on province and region codes. dataset: https://www.kaggle.com/sudalairajkumar/covid19-in-italy.

I have split the data, but when I try to set up the model, I get the following error:

XGBoostError: [16:16:15] C:/Users/Administrator/workspace/xgboost- 
win64_release_1.0.0/src/objective/multiclass_obj.cu:115: 
SoftmaxMultiClassObj: label must be in [0, num_class).

Code is as follows:

train = df[['RegionCode','ProvinceCode']].astype(int)
test = df['TotalPositiveCases'].astype(int)
X_test, X_train, y_test, y_train = train_test_split(train, test, 
test_size=0.30, random_state=42)

train = xgb.DMatrix(X_train, label=y_train)
test = xgb.DMatrix(X_test, label=y_test)

param = {
'max_depth':4,
'eta':0.3,
'objective': 'multi:softmax',
'num_class': 3}
epochs = 10

model = xgb.train(param, train, epochs)

the model attribute is where I get the error

Answer 1

This error comes when there are more labels in your target features than mentioned in the num_class parameter.

You should check if your target has more features than your num_class parameters or what you can do is to print the target.unique() as there might be some nulls or NAN in the data.