Mirror of forum.nim-lang.org

13242 :: I wrote a pandas-like library for data science in Nim

[2025-07-23T00:00:51+02:00]

eternal (orginal) [2025-07-23T00:00:51+02:00] view original

Repo: https://github.com/navid-m/razor

The goal at the end is to have near 1:1 functional similarity and ease of use as Pandas, though it'll probably be leaner.

oyster (orginal) [2025-07-23T04:17:37+02:00] view original

is possible to switch the license to a commercial or closed_sourced friendly one?

emre (orginal) [2025-07-23T04:23:46+02:00] view original

People are eventually realizing that Pandas does not have a good API. If I were creating a new library, I'd copy the Polars or Fugue API.

Integrations to consider: DataFusion, Arrow, and Lance.

eternal (orginal) [2025-07-24T13:10:58+02:00] view original

no

reallarry (orginal) [2025-07-24T13:14:04+02:00] view original

Maybe similar to: https://github.com/Luteva-ssh/nivot While nivot is focussed on pivot table analysis.

Clonk (orginal) [2025-07-24T15:16:42+02:00] view original

On datafusion I've started minimal bindings to C API of the Rust implementation https://github.com/SciNim/datafusion

It's untested and will progress very slowly since I don't have as much time as I'd like but PR will be reviewed and merged A

oyster (orginal) [2025-07-25T04:26:44+02:00] view original

what is the license of https://github.com/SciNim/datafusion ?

alexeypetrushin (orginal) [2025-07-25T05:01:07+02:00] view original
DataFrame has no validation (type safety) on column names. Typed structure may be better choice.
```
type
  StockReturns = object
    returns: seq[float]
    symbols: seq[string]
```