Duplicate rownames in scores were not handled properly before when
subsets occur since the input data is assumed to be a data.frame where
duplicates are impossible. However, a numeri matrix can handle duplicate
rownames. In the case where duplicates are possible the output defers to
the subset argument to determine which values are NA in the score output
instead of matching the rownames. Unit tests added. Running the tests on
current master will show the case where reconstruction of scores from subsets
fails.
Coverage increased (+0.03%) to 85.326% when pulling d5c26d7df27d7397c46536936a3b265fc84fff27 on DS-2841-2 into 65462b9f821f4ec9fd7f8f3eefcb8e6c6dfd46ea on master.
Duplicate rownames in scores were not handled properly before when subsets occur since the input data is assumed to be a data.frame where duplicates are impossible. However, a numeri matrix can handle duplicate rownames. In the case where duplicates are possible the output defers to the subset argument to determine which values are NA in the score output instead of matching the rownames. Unit tests added. Running the tests on current master will show the case where reconstruction of scores from subsets fails.