was trying to hunt down a "bug" (or at least fishy results) in the lead scoring model... started being suspicious of the tests at some point and i introduced much more realistic censoring
turns out everything works, but at least tests are a little bit stricter... it also surfaced some really weird fragility with the gamma regression model
Coverage remained the same at 67.385% when pulling 0ff68f450b374a78cae845d1b2d16a9b69fa6cae on test-censoring into 86a1718388295f7a4943bb3c60a35836db4233af on master.
was trying to hunt down a "bug" (or at least fishy results) in the lead scoring model... started being suspicious of the tests at some point and i introduced much more realistic censoring
turns out everything works, but at least tests are a little bit stricter... it also surfaced some really weird fragility with the gamma regression model