Closed Schwarzam closed 3 months ago
@Schwarzam This solutions that you propose sounds good. Would you like to contribute it to the codebase? We made you a member of hipscat-friends group which should make it possible for you to contribute to the github directly.
Feature request or suggestion
The current implementation of converting FITS data to pandas DataFrames using astropy.table.Table.to_pandas() results in string columns being returned as byte strings. Example:
This could be solved by adding this on FitsReader.read():
I ran some tests to check the performance on some ways to decode 2 columns with 10mi objects.
This is what I got on my M1.
Time taken by method_stack_unstack: 7.13616 seconds Time taken by method_direct_apply: 2.56532 seconds Time taken by method_vectorized_str_decode: 2.86572 seconds Time taken by method_list_comprehension: 2.42810 seconds