(Super simple) Summarisation and Plotting

Introduction

sumplot provides some very simple functionality to discretise, summarise and plot data.

The example below uses the diabetes dataset and summarises the variables s1, s2 and s3 by bp. The bp variable is discretised into buckets with a minimum of 5% of weight per bucket.

import polars as pl
from sklearn.datasets import load_diabetes

from sumnplot.discretisation import EqualWeightDiscretiser
from sumnplot.summarisation import group_by_weighted_average, ResponseWeight
from sumnplot.plot.altair.one_way_summary import produce_one_way_summary_plot

X, _ = load_diabetes(return_X_y=True, as_frame=True)
df = pl.DataFrame(X).with_columns(pl.lit(1).alias("w"))

bp_discretiser = EqualWeightDiscretiser(
    column="bp",
    weights="w",
    min_weight_proportion=0.05,
    round_breaks_for_labels=True,
)
bp_discretiser.fit(df)
df = bp_discretiser.transform(df)

results = group_by_weighted_average(
    df,
    groupby_columns=["bp"],
    responses=[ResponseWeight("s1", "w"), ResponseWeight("s2", "w"), ResponseWeight("s3", "w")],
)

chart = produce_one_way_summary_plot(
    results,
    x_axis_column="bp",
    left_y_axis_column="w",
    right_y_axis_columns=["s1", "s2", "s3"],
    title="One Way Summary by 'bp'",
    x_axis_label_angle=90,
)

Install

The easiest way to get sumnplot is directly from pypi using:

pip install sumnplot

Documentation

Documentation can be found at readthedocs.

For information on how to build the documentation locally see the docs README.

Build

sumnplot uses uv as the project management tool.

To install sumnplot for development, first install uv then run the following from the project root:

uv sync