openml / OpenML

Open Machine Learning
https://openml.org
BSD 3-Clause "New" or "Revised" License
664 stars 90 forks source link

Unable to fetch dataset qualities VIA API for a number of datasets #915

Open sarim-zafar opened 5 years ago

sarim-zafar commented 5 years ago

For the following datasets 252,1092,4537 I am encountering the following problem. When I try to fetch dataset qualities via the https://www.openml.org/api/v1/json/data/qualities/ API {"error": {"code": "364", "message": "Dataset processed with error", "additional_information": "Read timed out"}}

sarim-zafar commented 5 years ago

@joaquinvanschoren @janvanrijn can you guys please look into this. Plus I couldn't find the code for all the features in one place. I'd like to help in that area if possible.

janvanrijn commented 5 years ago

Some datasets have internal problems, and can qualities can therefore not be computer. That is indicated by this error.

The reason that these datasets have been processed with an error is the result of another open issue, can't find the exact number.

joaquinvanschoren commented 5 years ago

951 has been fixed

Dataset 252 is fine now Dataset 1092 had a weird error 'Error: Index 599 out of bounds for length 1' - but the ARFF file looks OK. Maybe an evaluation engine issue? Dataset 4537 has obvious errors in the XML description.

Should keep datasets 'active' if they have processing errors?

azrdev commented 5 years ago

When I try to fetch dataset qualities via the https://www.openml.org/api/v1/json/data/qualities/ API {"error": {"code": "364", "message": "Dataset processed with error", "additional_information": "Read timed out"}}

I'm getting this error too for https://www.openml.org/api/v1/json/data/qualities/8, the liver-disorders dataset from UCI

Edit: it's a 364 but a different message. another issue, or fits in here fine?

joaquinvanschoren commented 5 years ago

Aha, this is a known issue and I actually already submitted a PR: https://github.com/openml/EvaluationEngine/pull/30

@janvanrijn can you please merge and re-deploy the evaluation engine? Thanks!