using other data sets - pandas data frame about causaldiscoverytoolbox

[diviyank]

Hi,
no need, just feed the pandas dataframe to the functions :)
Best,
Diviyan

[andrewczgithub]

[andrewczgithub]

Hi @Diviyan-Kalainathan
Hope your well.
does this result mean this is no cause and effect.
Using a very small dataset of 10 values and two variables just to test the algorithms?

Many thanks,
Best,
Andrew

[diviyank]

Hi,
Actually, the CGNN algorithm only searches in the graph space of the provided skeleton. I suppose that your skeleton is empty (not enough dependence between variables), that might come from the fact that you have only 10 values : the p-value associated with the dependence score is high, making the edge not retained.
I order to have more consistent results, I would suggest using datasets of at least 100 samples.
And using the pairwise algorithms for datasets of 2 variables.

Best,
Diviyan

[andrewczgithub]

Thank you @Diviyan-Kalainathan !!
Much appreciated