Through previous PR (namely #267), the handling of near zero features in multiblock contexts seems quite lacking. There are many areas in which near zero variance feature filtration is applied when it shouldnt, and not applied when it should.
Additionally, the warnings provided to the use are cryptic and don't notify them of the issue. Users can only be made aware of nzv features in their data if they set the near.zero.var parameter to TRUE (or use the nearZeroVar() function). However, many new users miss these
Through previous PR (namely #267), the handling of near zero features in multiblock contexts seems quite lacking. There are many areas in which near zero variance feature filtration is applied when it shouldnt, and not applied when it should.
Additionally, the warnings provided to the use are cryptic and don't notify them of the issue. Users can only be made aware of nzv features in their data if they set the
near.zero.var
parameter toTRUE
(or use thenearZeroVar()
function). However, many new users miss these