Rye Tables vs Python/Pandas: A Different Way to Wrangle Data

Posted by symbolicard@reddit | programming | View on Reddit | 5 comments

[-]

middayc@reddit

Hi, I'm the author of the blog post / Ryelang. Any feedback is appreciated.

[-]

ROFLLOLSTER@reddit

I'm already switching most of my data analysis from pandas to polars, so I'd personally need a pretty compelling reason to use anything else.

Polars having a lazy API, query optimiser, and just generally being incredibly fast is one of its main selling points for me.

[-]

Table is one of Rye's base datatypes and it can't compete, at least performance-wise to specialized libraries like Polars. We could try to optimize it as much as possible, but this is not the main focus at this stage.

So, as it's written at the end of the blogpost, this is no a practical alternative you cn use instead of Python + Pandas / Polars today, but a show of concepts of Rye language.

Thanks for the reference to Polars, I will be looking at it for inspiration also.

[-]

PurepointDog@reddit

Look at the Polars lazy API. It's extremely well-suited for future optimizations, and they're actually just getting started implementing some of the more advanced query-planning and streaming optimizations.

Look at whether you could set something similar up so that you can expand/optimize in the future without needing a whole API rework

[-]

middayc@reddit

Thanks. I will look at it. You are the second or third today that is mentioning Polars. It is supposed to also have more immutable / endomorphic approach (as I've been told today on lobsters) so it seems the core philosophy is very similar.