Calculating xhat with ncomp=1 now works (#2)
nipals() now returns center and scale in the fitted model object, which makes it easier to get predictions from the model (#1).
New (experimental) function empca() to calculate principal components via weighted EM PCA.
Row/column names added to fitted matrix.
New startcol argument can now be a function which is applied to every column.
There is a slight change to automatic start column selection. By default, the start column is the column with the largest sum of absolute values. (Formerly was largest variance, but this does not make sense when columns are scaled.)
Row/column names added to fitted matrix.
New startcol argument can now be a function which is applied to every column.
There is a slight change to automatic start column selection. By default, the start column is the column with the largest sum of absolute values. (Formerly, the start column was the column with the largest variance, but this does not make sense when columns are scaled.)
nipals() function was split off from the gge package, extensively optimized and compared with implementations in other packages.