Make some fixes to handle data with missing values better:
Define as a constant column a column where there is only one non-missing value.
Change to NULL the value returned by rgcca_bootstrap_k for a bootstrap sample with a constant column in at least one block.
Add some functionalities for the correlation circle plots:
Add a new argument, display_blocks, which allows choosing which variables are plotted against the components of the selected block used to draw the correlation circle.
Forbid drawing a correlation circle with non-orthogonal components. If there is a superblock, the components of the superblock are used, whatever the original value of the block argument. If there is a response block, and block corresponds to this response block, an error is thrown.
Additional fixes:
Remove rounding of tuning parameters and change the strategy to replace long parameter combinations with set numbers.
Change rgcca_permutation to make the comparisons on the same permutations.
Make some fixes to handle data with missing values better:
NULLthe value returned byrgcca_bootstrap_kfor a bootstrap sample with a constant column in at least one block.Add some functionalities for the correlation circle plots:
display_blocks, which allows choosing which variables are plotted against the components of the selected block used to draw the correlation circle.blockargument. If there is a response block, andblockcorresponds to this response block, an error is thrown.Additional fixes:
rgcca_permutationto make the comparisons on the same permutations.