Data access performance tips

Hi,

Need some tips please. I need fast performance when:

1) Reading datafiles saved to disk

2) Accessing and modifying data when loaded

On reading data files I did some tests. Surprisingly parquet file is the slowest, .mat second and .csv fastest. Tried uncompressed .mat file as well. Data is saved as a table.

Are there any other recommendations? Should I split the numeric and text data and save them separately?

On accessing data when loaded - have had lackluster experience using table even though convenient. I'm assuming I should separate numeric and text data then create separate arrays.

Any other tips?

Have ~1 million rows with about 20 columns and want the access and calc to be fast.

Thanks