For XDF riskchart() fails if predictions has NAs

[Graham Williams]

Loading weatherAUS.xdf, build decision tree and Evaluate -> Risk Chart results in

Error in riskchart(crs$pr, crs$xdf.split2$RainTomorrow, crs$xdf.split2$RISK_MM, : Risk charts are for binary classification models only.

Reviewing the value of crs$xdf.split2$RainTomorrow showns that there are NA's. Is this confusing riskchart()? Might be a fix require for riskchart()

[Graham Williams]

Testing does find that riskchart() tests length(unique(ac)) and when ac has NA's that's an extra element. Need to handle NA's.