Closed erikbern closed 6 years ago
turns out the rate can also vary across groups and you can't fit a global one
This now fits the kaplan-meier estimates reasonably well, although the regression model itself seems kind of wacky:
i haven't figured out how to do uncertainty estimates with this yet, it's a bit more math involved but hoping to finish that at some point
ok mostly got uncertainty working... something with plotting timeseries is still broken, but i'm tempted to remove that one anyway
Coverage decreased (-9.8%) to 66.759% when pulling fff5f01f0557909cff060b5f57e659ffe3bec7bb on fit-lambda into 917d075a6024a07dcc8ff4b27f9db5952c906c4e on master.
turns out the rate can also vary across groups and you can't fit a global one
This now fits the kaplan-meier estimates reasonably well, although the regression model itself seems kind of wacky:
i haven't figured out how to do uncertainty estimates with this yet, it's a bit more math involved but hoping to finish that at some point