I work in a small scientific team, and we are facing issues with naming conventions for NumPy arrays. For example, consider the array name rawdata_freq_dist_time. This array contains raw data (rawdata), and its dimensions are freq, dist, and time. Therefore, it is a 3D array where the first dimension represents frequencies, and so on. The problem with this naming convention is that it has limitations, and we are considering a cleaner approach. I proposed using a data structure: from typing import NamedTuple import numpy as np class RawData(NamedTuple): data: np.ndarray freq_size: int dist_size: int time_size: int However, if I instantiate the class like this: rawData = RawData(data=np.zeros((128, 256, 12)), freq_size=128, dist_size=256, time_size=12) and then work with this structure: rawData.data = ... I lose information about the dimensions of my array. I cannot easily tell if it is 2D, 3D, etc., unlike with the name rawdata_freq_dist_time. I must admit, I am a bit stuck. How do your teams handle NumPy arrays to work efficiently and keep track of dimensions without getting lost? Thank you for your insights. Continue reading...
