I was checking the performance of write_dataframe and noticed there was a significant part spent in __ne__ of the geometry array, which can be optimized by using a built-in method for checking missing values instead of using != None (for a larger dataset this can take more than a second, while the notna() call is in the milliseconds)
I was checking the performance of
write_dataframe
and noticed there was a significant part spent in__ne__
of the geometry array, which can be optimized by using a built-in method for checking missing values instead of using!= None
(for a larger dataset this can take more than a second, while thenotna()
call is in the milliseconds)