+1 vote

Is there a way to see the full dataset of the predicted data of a give model ? Right know it seems to be limited to only 50k rows, and with my filters on I only get to see a very small portion of the rows I would like to study in order to enhance my model.

Thanks !
asked by

1 Answer

0 votes
Best answer

Hi Jérémie,

The recommended way to get predictions for a full dataset is to

  • deploy your best model to the flow
  • use it to score a dataset (in your case, I guess it is the same dataset as the one used for training.

See http://learn.dataiku.com/tutorials/103/part2.html for a detailed explanation.


Alternatively, in models → settings → Train & validation, you may change the dataset used for validation and get a report on the whole dataset. But beware that using the same data for training and validation yields artificially good performance, that will not generalize when predicting on new data.


I hope this helps,

answered by
edited by
861 questions
891 answers
1,162 users