PDAL uses structured numpy arrays, and so a dataframe must be casted to it by calling to_records(). It would be convenient for PDAL to consume dataframes directly to save the external call and allow people to not worry about transforming back and forth.
Add support for dataframes in Pipeline constructor when passing in data
Add a .dataframes member to provide convenient access to the data frame instead of numpy array
Don't copy data while doing it 😉
Make it optionally available depending upon whether or not pandas is importable
hobu
commented
1 year ago
PDAL uses structured numpy arrays, and so a dataframe must be casted to it by calling
to_records(). It would be convenient for PDAL to consume dataframes directly to save the external call and allow people to not worry about transforming back and forth.Pipelineconstructor when passing in data.dataframesmember to provide convenient access to the data frame instead of numpy array