hmbp 0.3.0

Simple matplotlib and PGFPlots plotting with consistent, publication-ready styling

hmbp

A simple matplotlib wrapper with consistent, publication-ready styling.

Installation

pip install hmbp

Style

• Font: Helvetica
• Colormap: RdPu (primary), PiYG (secondary/diverging)
• 400 DPI output
• Consistent sizing: title 15pt, labels 14pt, ticks 13pt, legend 12pt

Usage

import hmbp

fig, ax = hmbp.new_figure()
hmbp.line_plot(y_values, x_values, label="Model A")
hmbp.set_labels("Title", "X Label", "Y Label")
hmbp.save("output.png")

Available Functions

Function	Description
line_plot	Line with optional fill and smoothing
multi_line_plot	Multiple lines on same axes with smoothing
scatter_plot	Scatter with optional color mapping
histogram	Color-mapped histogram
histogram_overlay	Overlaid histograms for comparison
bar_plot	Vertical/horizontal bars
box_plot	Box plot distributions
violin_plot	Violin plot distributions
heatmap	2D heatmap with colorbar
line_plot_with_error	Line with shaded error region
confusion_matrix	Annotated confusion matrix
roc_curve	ROC curve with AUC
precision_recall_curve	PR curve with AP
residual_plot	Regression residuals
learning_curve	Train/val learning curves
metric_comparison	Horizontal bar comparison
volcano_plot	Volcano plot for differential analysis

Smoothing

Line plots support EMA smoothing for noisy data (e.g., training curves):

# smooth=0.9 means 90% weight on previous value (heavy smoothing)
hmbp.quick.line(noisy_loss, smooth=0.9, path="smoothed.png")
hmbp.quick.lines([y1, y2], labels=["A", "B"], smooth=0.8, path="comparison.png")

Helpers

• new_figure(figsize) - Create figure and axes
• set_labels(title, xlabel, ylabel) - Apply labels
• save(path, fig, close) - Save with auto-legend

Quick API

Single-call functions that create, label, and save in one step:

import hmbp

hmbp.quick.histogram(data, title="Scores", xlabel="Value", path="hist.png")
hmbp.quick.bar(values, labels, title="Comparison", ylabel="F1", path="bars.png")
hmbp.quick.confusion_matrix(cm, class_names=["A", "B"], path="cm.png")

Available: hmbp.quick.line, hmbp.quick.lines, hmbp.quick.scatter, hmbp.quick.histogram, hmbp.quick.histogram_overlay, hmbp.quick.bar, hmbp.quick.heatmap, hmbp.quick.confusion_matrix, hmbp.quick.roc, hmbp.quick.volcano, hmbp.quick.violin

PGFPlots Output

Generate LaTeX figures using PGFPlots for direct inclusion in papers:

import hmbp.pgfplots as pgf

# Step-by-step
pgf.new_figure()
pgf.line_plot(y, x=x, label="Data", marker="*")
pgf.set_labels(title="Results", xlabel="Time", ylabel="Value")
pgf.save("figure.tex")  # Creates .tex and .pdf

# Quick API
pgf.quick_bar(values, labels, title="Comparison", path="bars.tex")
pgf.quick_histogram(data, title="Distribution", path="hist.tex")

Features:

• Outputs both .tex and compiled .pdf files
• Helvetica font, clean axis styling, y-major grids
• Supports line_plot, multi_line_plot, bar_plot, histogram
• Log scale via new_figure(ymode='log')

Requires a LaTeX distribution with lualatex or pdflatex.