Polars Support¶

rusket accepts polars.DataFrame natively alongside pandas, via the Arrow-backed zero-copy path.

Installation¶

Install rusket with the Polars extra:

pip install "rusket[polars]"

uv add "rusket[polars]"

This pins polars>=0.20. If you already have Polars installed, you can also just pip install rusket.

Usage¶

The fpgrowth function detects Polars DataFrames automatically — no extra parameters needed:

import polars as pl
from rusket import FPGrowth

df = pl.DataFrame({
    "milk":  [True, True,  False, True],
    "bread": [True, False, True,  True],
    "eggs":  [False, True, True,  True],
})

model = FPGrowth(df, min_support=0.5)
freq = model.mine(use_colnames=True)
rules = model.association_rules(metric="lift", min_threshold=1.0)

How it works¶

The Polars path uses polars.DataFrame.to_numpy() which returns an Arrow-backed NumPy buffer — zero-copy for numeric dtypes.

Polars DataFrame
    │
    ▼  df.to_numpy()  (zero-copy for bool/int dtypes)
numpy uint8 array
    │
    ▼  FPGrowth.from_transactions(...)  (Rust, PyO3 ReadonlyArray2<u8>)
Rust FP-Tree / Eclat mining
    │
    ▼
pandas DataFrame  [support, itemsets]

Supported dtypes¶

freq = FPGrowth(lazy_df.collect(), min_support=0.3).mine(use_colnames=True)