[FIXED] TabNetRegressor not working with reshaped data

Issue

I am using the PyTorch implementation of tabnet and cannot figure out why I'm still getting this error. I import the data to a dataframe, I use this function to get my X, and y then my train-test split

def get_X_y(df):
    ''' This function takes in a dataframe and splits it into the X and y variables
    '''
    X = df.drop(['is_goal'], axis=1)
    y = df.is_goal
    
    return X,y

X,y = get_X_y(df)

from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.25, random_state=101)

Then I use this to reshape my y_train

y_train.values.reshape(-1,1)

Then create an instance of the model and try to fit it

reg = TabNetRegressor()
reg.fit(X_train, y_train)

and I get this error

ValueError: Targets should be 2D : (n_samples, n_regression) but y_train.shape=(639912,) given.
Use reshape(-1, 1) for single regression.

I understand why I need to reshape it as this is pretty common, but I cannot understand why it's still giving me this error. I've restarted the kernel in notebooks so I don't think it's persistence memory issues either.

Solution

You have to re-assign it:

y_train = y_train.values.reshape(-1,1)

Otherwise, it won't change.

Answered By - Berriel