We had discussed engineering features for number of exclamation points, and number (or percent) of words in all caps. This needs to be done before vectorization (which removes punctuation and capitalization), but can be done for each review then added at some point to the vectorization pickle that is already completed.
We had discussed engineering features for number of exclamation points, and number (or percent) of words in all caps. This needs to be done before vectorization (which removes punctuation and capitalization), but can be done for each review then added at some point to the vectorization pickle that is already completed.