Reorder scaling, adding superblock and handling NA values operations.
Before this pull request, scaling was done first, then the superblock was set and finally NA values were handled. As the implemented methods for handling NA values were nipals and complete, scaling before or after handling missing values led to the same results. However, it led to have sometimes different number of rows between call$blocks and out$A or out$Y. In order to prevent this and to ease the addition of new methods for handling missing values, this PR makes handling NA happen before scaling.
In order to do so, rgccaNa.R and sgccaNa.R are removed and NA handling is now done in handle_NA.R. Therefore, rgccad and sgcca are directly called from rgcca. That is why arguments A and C are respectively renamed to blocks and connection.
Reorder scaling, adding superblock and handling NA values operations.
Before this pull request, scaling was done first, then the superblock was set and finally NA values were handled. As the implemented methods for handling NA values were
nipalsandcomplete, scaling before or after handling missing values led to the same results. However, it led to have sometimes different number of rows betweencall$blocksandout$Aorout$Y. In order to prevent this and to ease the addition of new methods for handling missing values, this PR makes handling NA happen before scaling.In order to do so,
rgccaNa.RandsgccaNa.Rare removed and NA handling is now done inhandle_NA.R. Therefore,rgccadandsgccaare directly called fromrgcca. That is why argumentsAandCare respectively renamed toblocksandconnection.