search

FenTechSolutions / CausalDiscoveryToolbox

Package for causal inference in graphs and in the pairwise settings. Tools for graph structure recovery and dependencies are included.
https://fentechsolutions.github.io/CausalDiscoveryToolbox/html/index.html
MIT License
1.08k stars 198 forks source link

One missing data pair for tuebingen #113

Closed zhangtj1996 closed 2 years ago

zhangtj1996 commented 2 years ago

Hi there,

I found there is one dataset missing when I load the tuebingen dataset. The total number should be 100 instead of 99 for version 1.0 in https://webdav.tuebingen.mpg.de/cause-effect/.

Could anybody have a check and update the dataset?

Best wishes.

diviyank commented 2 years ago

Hello ! It's not really a mistake : there is actually one multivariate pair in the Tuebingen dataset which format is not managed by most algorithms. Thus this pair has been knowingly ignored... :roll_eyes:

Best regards, Diviyan

zhangtj1996 commented 2 years ago

Thank you very much, then this issue could be closed.

diviyank commented 2 years ago

Great !

amir-rahnama commented 1 year ago

There are 108 pairs of datasets in Tuebingen datasets, not 100. It is impossible to know which one is included in your dataset since you have started to call them differently than the original pairs (pair1 instead of pair0001).

In addition, you have reversed all the X and Ys so that all targets are now 1 (X causes Y). This makes it impossible to use AUC, for example, to measure the performance of these datasets.

I think updating the dataset with the right key from the original dataset, plus not reverting the columns but keeping the target as 0 will make your dataset much more useful.