Number parser performance can be an issue for large files. Some options: - Use numpy's own parser `np.asarray(['123', '23', 'nan'], dtype=float)` - [Fastnumbers](https://github.com/SethMMorton/fastnumbers), though it's an additional dependency.