Error while implementing the Naive Bayes approach to generate sentiment scores for movie reviews in NLPIA

[SallyAfua]

Hello NLPIA readers. Please I tried to implement the code snippet on generating sentiment scores for movie reviews in the NLPIA.

This line of code

'' movies['predicted_sentiment'] = nb.predict_proba(df_bows) * 8 - 4 ''

gave me this error

''Wrong number of items passed 2, placement implies 1''

Maybe the error is trying to say that I can't fix a two dimensional array into a dataframe's column, but I don't know how to fix it. Please can someone be of help. Thank you.

[jdiego]

Hi @SallyAfua the correct code is the following:

movies['predicted_sentiment'] = nb.predict_proba(df_bows)[:, 1] * 8 - 4

Because the predict_proba returns an array-like of shape (n_samples, n_classes). So, this method returns the probability of the samples for each class in the model.

Best Regards