Handling daily records of employees in technician attrition problem

[jasveen23]

I have daily data of active and terminated employees for the last 4 years. I used CoxPH model for regression and used cluster_col with the ID column as there were around 50 records per technician.

1. For the test data, I again have multiple records. While predicting the remaining time using predict_median function, do I need to indicate ID column again to get one prediction per employee?

2. It is basically a time series data with daily records such as number of tasks assigned, tasks completed, hours taken etc. and other HR data such as age, location.
   Is there any other model that would be more suitable to incorporate the concept of time?

3. Is there a method of evaluation on validation data except C index and k fold validation?

It would be really helpful if someone could help with these questions. Fairly new to survival analysis and lifelines package.
Thank you!