Strange calibration plot

[MuradOmarov]

Hello!
I'm trying to use lifelines in my study. I want to calibrate Cox model (with 10 predictors) with survival_probability_calibration.
There are about 2 000 cases and 130 000 controls in my dataset.
Here my event table from KM for better understanding:

And cumulative density:

I got C-index 0,77 after fitting. But Calibration curves look very strange:


Can you tell me what I'm doing wrong?
Thanks in advance

[CamDavidsonPilon]

Hi there, I don't think you are doing anything wrong. The error is likely in the calibration part - there may just be too few cases to build a reliable calibration model. Are you able to share with me the dataset? Publically, here, or privately, at cam.davidson.pilon@gmail.com

[MuradOmarov]

Thanks for your reply!
I sent the data to the e-mail.