kra

A set of useful tools to work with polars, providing convenient extensions for DataFrame manipulation, column operations, label encoding, and more.

Installation

Install kra from PyPI using pip:

pip install kra

To build and install a local version for development or testing:

pip install build
python -m build
pip install dist/kra-*.whl

This will build the wheel and install it into your current environment.

Features

• DataFrame and Series extensions: Add new methods to polars DataFrames and Series.
• Column utilities: Easily rename, check, and transform DataFrame columns.
• Label encoding: Encode string labels as categorical/integer values.
• Dict-of-dicts conversion: Convert between DataFrames and nested dictionaries.

---

Example Use Cases

1. Dict-of-Dicts Conversion

Convert a DataFrame to a dict of dicts using a column as the key:

import polars as pl
import kra

df = pl.DataFrame({
    "id": [1, 2, 3],
    "name": ["Alice", "Bob", "Charlie"]
})

dod = df.to_dod("id")
# {1: {'id': 1, 'name': 'Alice'}, 2: {'id': 2, 'name': 'Bob'}, ...}

# Convert back:
df2 = kra.from_dod(dod, "id")

---

2. Column Name Transformations

Transform column names to different cases:

import polars as pl
import kra

df = pl.DataFrame({
    "First Name": [1, 2],
    "Last Name": [3, 4]
})

df_lower = df.cols.to_lowercase()
df_camel = df.cols.to_camelcalse()
df_snake = df.cols.to_snakecase()

---

3. Label Encoding

Encode string labels as integers:

import polars as pl
import kra

df = pl.DataFrame({
    "label": ["cat", "dog", "cat", "bird"]
})

# Series API
encoded = df["label"].label.encode()

# Expression API (for use in with_columns, etc.)
df2 = df.with_columns(
    pl.col("label").label.encode().alias("encoded_label")
)

---

4. DataFrame Utilities

Drop columns of type Null:

import polars as pl
import kra

df = pl.DataFrame({
    "a": [1, 2, 3],
    "b": [None, None, None]
})

df_clean = df.drop_null_cols()

---

5. From Array-like

Create a DataFrame from a numpy array:

import kra
import numpy as np

data = np.array([[1, 2], [3, 4]])
df = kra.from_arraylike(data, schema=["x", "y"], orient="col")

---

---

API Reference

• kra.from_dod: Create DataFrame from dict of dicts.
• kra.to_dod: Convert DataFrame to dict of dicts.
• kra.Cols: DataFrame column utilities (access via df.cols).
• kra.LabelSeries: Series label encoding (access via series.label).
• kra.LabelExpr: Expression label encoding (access via pl.col(...).label).
• kra.drop_null_cols: Remove columns of type Null.
• kra.from_arraylike: Create DataFrame from array-like objects.

For more, see the intro.ipynb notebook.

---

kra includes a Rust extension for fast label encoding, accessible via the Python API.

License

MIT License. See LICENSE for details.