+2 votes
by

1 Answer

+7 votes
Best answer

There are three kinds of datasets in dataiku: 

  • Managed datasets:  datasets that are created by recipe. Dataiku assumes he “owns” the data and schema of those datasets
  • Uploaded datasets:  raw files that are uploaded through the user interface. They are actually stored locally in raw form, in a specific folder (that you can find in DATA_DIR) with a subfolder per dataset name PROJECTNAME.DATASETNAME
    You can “modify” an uploaded dataset by reuploading it. An uploaded dataset is actually a folder, meaning that it can contain several files
  • Non-managed, non-uploaded datasets. Usually it's a dataset that points to exisiting data (a table in a sql table for instance) that you can use as an input for a recipe 

 

by
selected by
Thank you!
1,052 questions
1,091 answers
1,196 comments
9,126 users

©Dataiku 2012-2018 - Privacy Policy