Using general functionalities on new docs

[VenkataBuddha]

After predicting topics on new docs using transform function. How can we update or make use of the visualization methods and other topic info methods on the new docs?

For ex:
topic_model.get_document_info(test['Body_title_abs_without_stopwords'])

WHEN I RUN THE ABOVE CODE: I get the following error: array length 5102 does not match index length 3364

The number 5102 is the length of my training docs and 3364 is the length of my test docs.

[MaartenGr]

The .get_document_info function is currently only used for documents on which the model was trained as information about unseen documents are not saved within the topic model and you would have to pass manually from the output variables.

[VenkataBuddha]

Thanks for the response!
Can you provide any sample example of what you meant by manually passing from output variables?

[MaartenGr]

In order to create similar output of .get_document_info, you would have to create the dataframe yourself. So let's say you did the following:

predicted_topics, predicted_probs = topic_model.transform(unseen_documents)

then you can create a similar dataframe as follows:

import pandas as pd
df = pd.DataFrame({"Document": unseen_docs, "Topic": predicted_topics, "Probability": predicted_probs})

If you would like to have more metadata, you can combine df with the output of .get_document_info.

[VenkataBuddha]

Is there a manual way to visualize the new documents?
like to generate a plot simialr to visualize_documents() method

[MaartenGr]

That is currently not implemented, you would have to adapt the code here in order to do that for unseen documents.