Question

Before, there was larry and structured/record arrays in NumPy, but I wonder if they are used any more with any frequency given the rapid development of the pandas package. Coming from R, I would always get stuck having to unpack the record arrays to modify values from multiple columns and reassign them back into the structure but I’m so glad that pandas now allows this for its data frames. I wonder if there are any uses for which record arrays are still superior (does it have some useful methods that pandas does not have)?

Answer 1

Here’s a good explanation and simple comparison between pandas and numpy record arrays – Normalize/Standardize a numpy recarray