0 votes
The dataset generated by prediction is empty. No errors are shown and the prediction job executes normally. Are there any possibilities to debug this?
Could you please attach a diagnostic of the affected job ? From the job page, click on Actions > Download job diagnosis.
If the resulting file is too large for mail (> 15 MB), you can use a file transfer service like WeTransfer to get it to us
Hi, Have you been able to solve your issue?
Well kind maybe.

Could it be that filter applied within the visual analysis, gets also deployed to the flow when deploying the model?
Yes, all steps you add to the script part of the visual analysis are also deployed to the flow.

1 Answer

0 votes

When designing a Machine Learning algorithm in the Visual Analysis environment of DSS, you can add steps in the "Script" tab of the analysis. All these steps are applied before the data reaches the feature preprocessing and the model training/scoring. They are also bundled when you deploy the model to the flow or to a real-time API endpoint. This is similar to the concept of scikit-learn Pipeline, if you are familiar with it. It is actually a powerful way to bundle data preparation, feature processing and model scoring in a real-time setting.


1,233 questions
1,204 answers
11,676 users

┬ęDataiku 2012-2018 - Privacy Policy