How to write DataFrame compatible features

Question

I recently switched from using numpy ndarrayto pandas' DataFramefor my data analysis needs. I noticed that numpy functions seem to accept DataFrameobjects instead ndarraywithout any problems. However, when I try to use many of my existing functions written to work with ndarray, they often fail when performing indexing, translation, etc. operations, and I am forced to transfer the base one ndarrayusing df.values.

Is there a standard way or set of recommendations for ensuring function compatibility with DataFrame? How to execute numpy functions for both types?

Answer 1

I resorted to digging in the numpy source code and found that many functions simply convert the input to ndarray, first using functions such as asarrayor asanyarray.

def numpyFunction(x, *args, **kwargs):
    x = np.asanyarray(x)
    ...