Algorithm implemented for the SyncPy library : https://github.com/syncpy/SyncPy
This algorithm compares the relationship that two dataset can share.Let X and Y be those datasets.It seeks out for linear projections of the two datasets such that the i^{th} projection of X is maximally correlated with the i^{th} projection of Y.
The two datasets must have the same number of rows (same sample size) but can have a number of columns(features) that differs.
For a more complete description of the algorithm see :
- David Weenink, Canonical correlation analysis
- Karl Stratos, A Hitchhiker’s Guide to PCA and CCA
- Alvin C. Rencher, Methods of Multivariate Analysis