Jamie,

you have to be very careful when you use dependent variables in stochastic chains to meet requirements of mathematical theories. To avoid problems, you can maintain the dependence using a mathematical function (here a related question: https://www.researchgate.net/post/Text_Classification_How_to_model_labels_that_are_co-related)

There is an implementation of WEKA in Python 3 from Peter Reutemann (Waikato university) -- https://github.com/fracpete/python-weka-wrapper3. Here is the mailing list http://weka.8497.n7.nabble.com/ where you can ask your question.

Best, Alexander

Would you need to have temporal correlation in each stochastic variable, or is it sufficient to keep the correlation intact between the stochastic variables? If yes to the latter it is possible to use copulas for generating correlated outputs from the N number of variables. This is straight forward in Matlab at least, where correlations between variables is obtained via corrcoef and correlated synthetic data can be generated by copularnd. As regards python, it seems to have a copulalib package which could perhaps be used.

Thanks Alexander for the caution and options.

And thanks Joakim! I think this is what I was looking for.

As for the temporal correlation, it would be good to have a temporal component to the variables. Do you think it would be sufficient to drive the most significant variable using a temporal method, e.g. Markov chain, and then use copula theory to derive conditions of the other variables from this stochastic result?

I have just made a quick scatter matrix of all the variables against each other, and placing a histogram where the same variables meet.

Dir is cloud motion direction at 4 different heights.

u is cloud speed at 4 different heights.

T, w and p are temperature, precipitable water and pressure.

cld is total cloudiness.

Do you think all these variables are contenders for the Copula method?

Or perhaps there should be 2 copulas, the first for T, w and p. The second for u, and dir. cld seems independent.

Thanks

Dr jemie is probably right.

Yes, there is definitely some correlation going in many of these plots, so I think that copulas could work to some degree of accuracy. As regards the exact setup, I do not really follow how the variables are supposed to tie together. Generally, copulas can handle many variables, so it could be easier to include even seemingly uncorrelated ones, just in case there might be some small correlation. But, then again, I do not really follow all variables here.