grizz

0.1.1 verified Mon Apr 27 auth: no python

A lightweight library for preprocessing data with Polars. Version 0.1.1 supports Python 3.9-3.13 and dependencies include Polars, PyArrow, and ClickHouse. Releases are frequent, adding transformers and temporal features.

pip install grizz

Common errors

error ModuleNotFoundError: No module named 'grizz.transformer.ts' ↓

cause The `ts` submodule was removed in v0.0.3 when temporal transformers were moved up.

fix

Use from grizz.transformer import ToDatetimeTransformer instead.

error TypeError: 'columns' must be a list of strings ↓

cause Passing a single string to `columns` instead of a list.

fix

Wrap the column name in a list: columns=['col_name'].

error polars.exceptions.ComputeError: conversion from `str` to `i64` failed ↓

cause Casting a non-numeric string to an integer type.

fix

Ensure all values in the column are numeric, or use FloatCastTransformer for floats.

Warnings

breaking In v0.0.5, the `period` parameter was renamed to `interval` in several transformers (e.g., CumSumTransformer, CumMeanTransformer). Code using `period` will break. ↓

fix Replace `period` with `interval` in transformer constructors.

breaking The `ToTimeTransformer` and `ToDatetimeTransformer` were moved from `grizz.transformer.ts` to `grizz.transformer` in v0.0.3. Old imports will fail. ↓

fix Update imports: `from grizz.transformer import ToTimeTransformer`.

gotcha Transformer base class methods `transform` and `fit_transform` expect Polars DataFrames. Passing pandas DataFrames will raise a TypeError. ↓

fix Convert pandas DataFrame to Polars using `pl.from_pandas(df)` before passing.

Imports

CastTransformer

wrong

from grizz import CastTransformer

correct

from grizz.transformer import CastTransformer

Transformer classes are in the 'transformer' submodule

InPlaceCastTransformer

wrong

from grizz.transformer.cast import InPlaceCastTransformer

correct

from grizz.transformer import InPlaceCastTransformer

Use top-level transformer import

ToDatetimeTransformer

wrong

from grizz.transformer.ts import ToDatetimeTransformer

correct

from grizz.transformer import ToDatetimeTransformer

Moved to core transformer module as of v0.0.3

Quickstart

Create a Polars DataFrame and apply a CastTransformer to cast column 'a' to Int64.

import polars as pl
from grizz.transformer import CastTransformer

df = pl.DataFrame({'a': ['1', '2'], 'b': ['3.0', '4.0']})
transformer = CastTransformer(columns=['a'], dtype=pl.Int64)
result = transformer.fit_transform(df)
print(result)