Combining Two Different models - One built on text attributes and one built on numeric attributes

BarryE

Hello,

I am working on a college research project and am looking for solutions to an issue I am facing. I want to build two seperate models, one utilising the text data and one utilising the numeric data. I plan on applying Text Preprocessing steps to the text model and traditional pre processing steps for the numeric model. However, I am unsure of how is the best way of combining the power of both these models into a single model and was hoping someone could kindly shed some light on it for me?

Thanks in advance for any responses.

Kind Regards,
Barry

yyhuang

Hi Barry,
If you are using "text vectorization" for text preprocessing, you can directly apply it by "Apply Model". Take a look at the tutorial process for text vectorization operator, which has applied both text preprocessing models and predictive models.

Of course, you can group models and apply the grouped model just once. See attached for details.

If you are using the "text processing" extension for text preprocessing, you will not have a "model", but you can re-use the text processing sub-process on any new documents with the wordlist from your training set.. The sub-process for text pre-processing step can be shared as building blocks, or even shipped as a custom operator.

HTH!

Cheers,
YY

BarryE

Hi YY,

Thank you so much for the quick response.

I will be using the Text Processing extension, so looks like I will be going down the path of using Group Models.

I may be looking for your expertise down the line

Thanks Again,
Barry

yyhuang

Again, if you are building sub-process with text processing extension, you will not be able to "group models". Instead, you can re-use the sub-process on new document. The "group models" is applicable only if you use "text vectorization" model for document pre-processing.