I’m currently playing around a bit with the Wanikani-API and I can already import all the data into an Access-Database. Yeah!
Now I want to do some analysis stuff with the data and found something in the data that doesn’t make sense to me. I hope someone knows the answer and can explain it to me:
For each review item in the collection I read the data_updated_at and the created_at dates and write them in my db table. Now I noticed that for about 10% of all the review records the data_updated_at is bigger than created_at. Sometime even around 30 days.
Now my question is: What’s the reason that a review record is getting updated? And why sometimes 4 weeks after the review has been created?
Normally review records don’t get updated after they have been created. But recently the WK devs have been backfilling the review data to align with API changes on the backend: they’ve added and removed attributes related to this big backend change:
The result is that the updated timestamp gets bumped, even though the essentials of the review result haven’t changed.
The records you are mentioning are probably from before this change, and have been touched by that process. And the more recent review records will most likely be untouched. Your code should allow for the possibility that these kinds of changes will happen occasionally.
Thank you for the answer! That could explain it. Most of the updated reviews are quite old… But I even have some from today that got updated already. But I will just ignore the updated timestamp then…
The updated timestamp is always filled in, even if the record has never been updated. Initially, the created timestamp and the updated timestamp will be identical. They will only drift apart if they’re updated after they’ve been created.
Sorry, I might have been not really clear. When I said, got updated I mean that the timestamp of “data_updated_at” is newer than the one in “created_at”.
I am.
So all reviews created until June 10th have and update-date that is between mid and end of june.
Most reviews created after june 10th, have identical update dates, but around 2-3% (estimated) are not. For most of them it’s just 1 second difference, which could be explainable. But for some it’s a couple of days…
Anyway. If the functional review data doesn’t get updated, I don’t need to worry about the updated timestamp…
