I have a text data column in my dataset and want to create embeddings using a library which might not be present in DataRobot, Do I have the freedom to manually transform the text data or any other data for that matter. Thanks in advance.
Solved! Go to Solution.
@anton.kasyanov, I would also note that Composable ML isnt currently available in enterprise by default either.
First of all, DataRobot’s built in capabilities in NLP are very strong. You can try to just use your original text dataset and see what kind of accuracy do you get.
If you still want to use your custom embeddings processing, you can either do it outside of DataRobot or leverage Composable ML ( see a demo here). Using Composable ML with AutoPilot is the recommended path, just note that Composable ML is not available in the AI Cloud Platform Trial yet.
DR does use word embedding techniques such as Word2Vec and fastText with both CBOW and Skip-Gram learning methods. If you wanted to use your own embeddings library I think you would have to include the embedding in your dataset before modelling in DR.