Write script to gather SAM reliabilty for client contemplating machine learning

[levithatcher]

Some non-hosted clients don't have the EDW/SAM infrastructure to a point where it can support a) ~99.9% uptime and b) early morning SAM.

This R script would roughly have these requirements (at first pass):

1. Use a clinical SAM that's already been in production for a few months at the client and which is refreshed nightly. For new clients this would obviously not work.

2. Use the SAM date/time stamp (on the summary table) to check the percentage of days in the last 1, 2, 3 months that the SAM has finished.

Would be fantastic to plot the average time in the morning that the SAM in question finished (to see how variable it's been). It'd be great to hear what the platform team is doing in this regard.