I have a dataset, that I use label_encoding for data preprocessing before feed into the model.

I first label_encoding all of my data, then split the data into training and testing.
And I want to use “label_encoder.inverse_transform” to reverse the code, but it has this error message:

y contains previously unseen labels: [8 9]

The error message was clear, but that’s the part that I am confused, because I label-encoded all of the data before splitting.

from sklearn.preprocessing import LabelEncoder
label_encoder = LabelEncoder()
category =[]
data_encoded = data_cleaned
for i in data_cleaned.columns:
    if data_cleaned[i].dtype == 'category':
        category.append(i)
        data_encoded[i] = label_encoder.fit_transform(data_cleaned[i])

Split the training and testing data

target_variable = ['target']
data_encoded = data_encoded.dropna(subset=target_variable)
x = data_encoded.drop(target_variable, axis=1)
y = data_encoded[target_variable]

And try to reverse the encoded label

//only want to reverse the category column
for i in category:
    X_test[i] = label_encoder.inverse_transform(X_test[i])
    print(X_test[i])

If anyone could point some direction that would be great!
Thank you!