James Chapman

OK the source of this is basically a misalignment between RCCA lambda and cca_zoo c. lambda gets added to the covariance matrices but c slides between 0 regularisation and 1...

Yeah definitely! Will almost certainly add this one in somewhere. Obviously there's some sharp edges (e.g. here where the parameterizations don't quite match) but no harm in having them

I've used variants of your script to check that the source of the difference is lambda/c and convinced myself. PRCCA with no regularisation (and fewer features than samples) everything matches...

I assume you mean you get that error when you run exactly the code in the example? If so could you share scikit-learn version as that's the only place I...

OK I've been made of this error already. I think it's outside my package and perhaps a problem with using python 3.6. Could you raise the issue on tensorly and...

Yeah it's already been raised here: https://github.com/tensorly/tensorly/issues/479 So need to use Python 3.7

Trying to understand this question but struggling a bit (perhaps a language barrier or perhaps terminology). **In principle the first dimensions of the learnt representation of X,Y,Z call them Zx,...

I'd be most immediately surprised if the first dimension in your training data did not have the highest average correlation. The other observations with respect to explained covariance are entirely...

Yeah I don't think it's _biasing the algorithm_ so much as it perhaps makes some part of the solving process unstable. The default behaviour is to apply PCA to the...

also just checking we are referring to training/in-sample correlations as opposed to testing/out-of-sample correlations. _anything_ is possible out of sample!