dfmp

A matplotlib-based library that generates bar charts where human silhouettes are horizontally scaled (wider/thinner) and vertically scaled (taller/shorter) to represent data values.

How it works

Each silhouette is generated by interpolating between pre-computed body profiles stored in silhouette_data.npz. The data values are min-max normalized to determine both the "fatness" (horizontal width) and height of each figure. The silhouettes are rendered as matplotlib.patches.PathPatch objects, so they integrate naturally with the rest of the matplotlib ecosystem.

Installation

Requires Python 3.9+.

pip install dfmp

Or install locally for development:

git clone https://github.com/djemec/dfmp.git
cd dfmp
pip install -e ".[dev]"

Dependencies

• matplotlib >= 3.5
• numpy >= 1.21
• scipy >= 1.9

Quick start

from dfmp import plot

plot(['A', 'B', 'C', 'D', 'E'], [12, 35, 28, 50, 19])

The API is designed to mirror plt.bar() -- if you know how to make a bar chart, you know how to make a dfmp plot.

API

plot(x, height, *, color=None, alpha=None, tick_label=None,
     label=None, show_values=True, value_format=None,
     ax=None, **kwargs)

Parameters

Parameter	Type	Default	Description
x	array-like	(required)	String labels or numeric positions. Strings are used as tick labels automatically.
height	array-like	(required)	Data values that drive both fatness and vertical scaling of each silhouette.
color	str or list	'#5B9A8B'	Single color for all figures, or a list of colors (one per figure).
alpha	float	None	Opacity (0--1) applied to every silhouette.
tick_label	list of str	None	Override x-axis labels when x is numeric.
label	str	None	Legend label for this set of silhouettes.
show_values	bool	True	Annotate each silhouette with its data value.
value_format	str or callable	None	Format string (e.g. '{:.1f}%') or callable f(val) -> str for value annotations.
ax	Axes	None	Target matplotlib axes. Creates a new figure if None.
**kwargs			Passed through to matplotlib.patches.PathPatch.

Returns

matplotlib.axes.Axes -- the axes containing the chart (following the seaborn convention).

Examples

from dfmp import plot
import matplotlib.pyplot as plt

ax = plot(
    ['A', 'B', 'C'],
    [10, 50, 100],
    color=['#E76F51', '#2A9D8F', '#264653'], #optional
    value_format='{:.1f}' # optional
)

#matplotlib config examples
plt.title('My chart')
plt.xlabel('Categories')
plt.show()

# save to a file
ax.figure.savefig('chart.png', dpi=150, bbox_inches='tight', facecolor='white')

Drawing on an existing axes

Pass an ax to draw on a pre-existing figure, just like seaborn or matplotlib.

import matplotlib.pyplot as plt
from dfmp import plot

fig, ax = plt.subplots(figsize=(10, 6))
plot(['A', 'B', 'C'], [10, 20, 30], ax=ax)

Testing

pytest --tb=short -q

With coverage:

pytest --cov --cov-report=term-missing

License

MIT