Greetings,   Hopefully I can be of some help! Proc optbinning is primarily used internally in the Interactive Grouping node within Model Studio, and thus information surrounding it can be a bit cryptic, as you’ve found. Maybe getting some information on how we prepare the data and use the proc will be beneficial.   Indeed as you’ve found, optbinning does not want missing values in the woeTrend column. I can confirm this - it is even how it was in pre-Viya SAS. It does seem a bit confusing based on what the column represents. I think as long as you have a sane value in that first bin, that corresponds to the monotonic trend of the woe across the bins you are fine. One way of handling this would be to use a regression to estimate the trend of the WOE across the bins, and use a predicted value for woeTrend for all bins (which would give you a sane value for that first bin). You would treat the woe as the dependent variable and use the lower bound of the bin or the bin number as the independent variable. (and 3). Unfortunately no there are no builtin procedures to facilitate building datain and/or parms. The most expected way of building these tables is to start with the output of something like proc binning which gives you the bin information, calculate the woe, and go from there. It seems like you are already doing this so you are on the right track. And w.r.t. bins for missing values, we eliminate these from datain before call the proc - I don't think proc optbinning can handle the missing value bin. Hope this helps!   ... View more