Default settings for importing datasets
So, I literally never want my integers coerced to doubles, and it would be great to have the option to have the "Normalize floats & doubles" box default to unchecked when I'm creating a new dataset. Maybe on a per-project basis? Maybe a Datasets section in the Project Settings?
Similarly, I basically always want my dates to be imported as strings, and then I can parse them later. Would be lovely to have an option to set a default for that, too.
What would actually be ideal is if there was a way - even if it couldn't be set as a default - to tell Dataiku not to try to auto-detect meanings on Dataset creation at all. It's often finding some number of integer-like entries in a foreign key column that actually do need to be text because some of them do have alpha characters in them, and it's often guessing that things are genders, or countries, that are nothing like that.
Thanks!
Comments
-
Turribeach Dataiku DSS Core Designer, Neuron, Dataiku DSS Adv Designer, Registered, Neuron 2023, Circle Member Posts: 2,722 Neuron"What would actually be ideal is if there was a way - even if it couldn't be set as a default - to tell Dataiku
notto try to auto-detect meanings on Dataset creationat all."This is available for Prepare recipes in Administration ⇒ Settings ⇒ Other ⇒ Misc:
-
Oh that's rad! Unfortunately, I cannot find it. When I go to [waffle menu]→Administration & Settings→Administration, the only sections are "License info" and "Features and quotas"… and my Project Settings page doesn't have any of the things yours does.
Is this likely an issue of the environment, or is there somewhere obvious I'm not looking?
Thanks! :)
-
Turribeach Dataiku DSS Core Designer, Neuron, Dataiku DSS Adv Designer, Registered, Neuron 2023, Circle Member Posts: 2,722 Neuron
Upgrade to the latest v14 version. You are probably running an older version.
-
Yeah we're on 14.7.2, and I still don't have all of that.
It really seems like it should be bog-standard to have an option, in the creation of a new dataset, to simply say "Please use the data types that correspond to what's coming out of the database, and nothing else, please and thank you", right? My columns are never a gender, and often not a US state or a country, and I so, so often have varchar fields where some of the rows have non-numbers but Dataiku says "Hey, this is definitely a number, boss (and since you missed the tiny checkbox this time, I'm going to add a '.0' to the end of it)!".
