using other data sets - pandas data frame about causaldiscoverytoolbox HOT 5 CLOSED

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

from causaldiscoverytoolbox.

from causaldiscoverytoolbox.

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

from causaldiscoverytoolbox.

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

from causaldiscoverytoolbox.

Thank you @Diviyan-Kalainathan !!
Much appreciated

from causaldiscoverytoolbox.