implementing np.nanmean producing unexpected results on masked arrays in NumPy 1.23
I'm maintaining legacy code that I'm working on a personal project and This might be a silly question, but I'm working with unexpected results when using `np.nanmean` on masked arrays in NumPy version 1.23..... I have a 2D array with some NaN values that I want to ignore in the mean calculation, but it seems like not all NaN values are being properly ignored, especially when there are masked values present in the same array. Here is a simplified version of my code: ```python import numpy as np # Create a masked array with NaNs data = np.array([[1, 2, 3], [4, np.nan, 6], [np.nan, np.nan, 9]]) masked_data = np.ma.masked_array(data, mask=[[0, 0, 0], [0, 0, 0], [1, 1, 0]]) # Calculate mean ignoring NaNs mean_value = np.nanmean(masked_data) print(mean_value) ``` This code produces `5.0`, which I expected to be the mean of `1, 2, 3, 4, 6, 9`. However, I believe it should account for the masked values correctly and return the mean of the non-masked, non-NaN values, which should actually be `4.0`. I also tried using `masked_data.compressed()` before passing it to `np.nanmean`: ```python mean_value_compressed = np.nanmean(masked_data.compressed()) print(mean_value_compressed) ``` This gave me `5.0` as well. Could someone explain why `np.nanmean` behaves this way with masked arrays? Is there a workaround to correctly compute the mean ignoring both NaNs and masked values? Any insights would be greatly appreciated! This is part of a larger web app I'm building. Has anyone else encountered this? I'd really appreciate any guidance on this. I'd really appreciate any guidance on this.