Discuss/request: required y survival status can be 0/1 integer in addition to bool about scikit-survival HOT 3 CLOSED

If somebody could post quick minimal example how you create this structured array from two lists, that would be very helpfu

I usually use numpy.fromiter as in this line.

It seems there are a couple of related issues,

1. Construction of structured array is not straight-forward.
2. Supporting passing a 2-column pandas DataFrame instead of structured array.
3. Allow for censoring indicator to be an array of 0/1 instead of boolean.

I believe it would be best to add a function equivalent to R's Surv that converts different inputs to the structured array format used internally.

from scikit-survival.

Absolutely for it! I think am quite capable in programming. But I still have no clue, how to get to the structured arrays and I've been trying for a while now! I know my python-skills are a bit rusty - but not this much.

If somebody could post quick minimal example how you create this structured array from two lists, that would be very helpful. Maybe add it to the example Python Notebook. Thanks

from scikit-survival.

@region-spotteR I use a pandas dataframe -> structured array workflow. Say you have pandas df with 2 columns: status and time. Convert status to bool and then convert the 2 columns into structured array

df.status = df.status.astype('bool')
df.to_records(index=False)

to recap, the point of this thread is to discuss/request that we don't need the first line

from scikit-survival.