forestplotx 1.0.1

Publication-ready forest plots for regression model outputs in Python.

forestplotx

Publication-ready forest plots for regression model outputs in Python.

forestplotx takes DataFrame output from logistic, linear, ordinal, or gamma regression models and produces a combined table + forest plot figure — ready for papers, reports, and presentations.

Features

• Multiple model types — binomial (logistic), linear, gamma, and ordinal (cumulative logit)
• Automatic effect-scale handling — exponentiation, log-scale axes, and reference lines driven by link function
• Flexible column detection — accepts OR, Ratio, Estimate, beta, Coef, or effect as input
• Dual-outcome layout — side-by-side comparison of up to two outcomes
• Category grouping — optional row grouping with bold category headers
• Deterministic layout presets — fixed internal geometry for 4 core display cases
• Adaptive small-table sizing — compact height heuristic for low row counts
• Static matplotlib output — high-resolution, saveable figures

Layout Examples

• examples/layout_case1_general_true_two_outcomes.png
• examples/layout_case2_general_true_one_outcome.png
• examples/layout_case3_general_false_two_outcomes.png
• examples/layout_case4_general_false_one_outcome.png

Installation

pip install forestplotx

Requires Python ≥ 3.10. Dependencies: matplotlib>=3.7, numpy>=1.24, pandas>=2.0.

Development install (reproducible environment)

pip install -r requirements.txt   # pin exact versions used during development
pip install -e ".[dev]"           # install forestplotx itself in editable mode

requirements.txt pins the full transitive closure of runtime + test dependencies. pyproject.toml declares the minimum-version constraints used when installing normally.

Quick Start

import pandas as pd
import forestplotx as fpx

# Example: logistic regression output
df = pd.DataFrame({
    "predictor": ["Age", "Sex", "BMI", "Smoking"],
    "outcome":   ["Mortality"] * 4,
    "Estimate":  [-0.12, 0.85, 0.30, 0.55],   # log-odds (pre-exponentiation)
    "CI_low":    [-0.35, 0.42, 0.05, 0.20],
    "CI_high":   [ 0.11, 1.28, 0.55, 0.90],
    "p_value":   [0.300, 0.001, 0.020, 0.003],
})

fig, axes = fpx.forest_plot(df, model_type="binom")

Supported Model Types

forestplotx supports multiple regression model families.
Effect interpretation and axis scaling are determined by the model family and link function.

model_type	Example models	Link	Effect label (table)	X-axis label	Reference line
"binom"	Logistic regression (glm, glmer)	logit	OR	Odds Ratio (log scale)	1.0
"gamma"	Gamma GLM / GLMM	log	Ratio	Ratio (log scale)	1.0
"linear"	Linear regression (lm, Gaussian GLM)	identity	β	β (coefficient)	0.0
"ordinal"	Ordinal regression (clm, clmm, polr)	logit	OR	Odds Ratio (log scale)	1.0

The link parameter can override the default — for example, model_type="binom", link="identity" will skip exponentiation and plot on a linear scale.

Interpretation Notes

• Binomial (logit) -> Odds Ratios (OR)
• Gamma (log link) -> Multiplicative mean ratios
• Linear (identity) -> Additive regression coefficients (β)
• Ordinal (logit) -> Cumulative Odds Ratios (OR)

Exponentiation is automatically applied for models using log or logit links.

Input DataFrame

Required columns

Column	Description
predictor	Row labels (predictor names)
outcome	Outcome name (used for column headers and filtering)
Effect column	One of: OR, Ratio, Estimate, beta, Coef, effect
CI_low / ci_low	Lower bound of 95% CI
CI_high / ci_high	Upper bound of 95% CI

Optional columns

Column	Description
p_value	P-value (bold formatting applied when < 0.05)
category	Group predictors under category headers
n	Event count
N	Total count

Note: For logit/log links, exponentiate=None applies model-based exponentiation with a warning; set exponentiate=False if your data is already on effect scale. Displayed CI values in the table use bracket notation: [low,high].

API Reference

forest_plot()

fig, axes = fpx.forest_plot(
    df,                              # DataFrame with model output
    outcomes=None,                   # list[str], max 2; auto-detected if None
    save=None,                       # File path to save (e.g. "plot.png")
    model_type="binom",              # "binom" | "gamma" | "linear" | "ordinal"
    link=None,                       # Override default link function
    exponentiate=None,               # None=auto by link, True=force, False=disable
    table_only=False,                # Render table without forest panel
    legend_labels=None,              # list[str] override for legend entries
    point_colors=None,               # list[str], up to 2 hex codes for outcome markers
    footer_text=None,                # Italic footer (wrapped/capped internally)
    tick_style="decimal",            # "decimal" or "power10" (readable log10 exponents)
    clip_outliers=False,             # Clip axis limits by quantiles (opt-in)
    clip_quantiles=(0.02, 0.98),     # Low/high quantiles used when clipping
    base_decimals=2,                 # Decimal places for effect / CI values
    show=True,                       # Call plt.show(); set False for programmatic use
    show_general_stats=True,         # Show n / N / Freq columns
    bold_override=None,              # Manual bold control per predictor/outcome
)

Returns: (fig, axes) — matplotlib Figure and axes tuple. When show=False, the figure is returned without displaying, allowing further customization before calling plt.show() manually. When exponentiate=None, auto exponentiation for log/logit links emits a warning so users can verify input scale.

Layout Behavior (v1)

forest_plot() uses fixed internal layout presets (including internal font size) for:

1. show_general_stats=True + two outcomes
2. show_general_stats=True + one outcome
3. show_general_stats=False + two outcomes
4. show_general_stats=False + one outcome

This is intentional to keep output stable and publication-ready across common use cases. base_decimals is capped at 3 internally to prevent table collisions in dense layouts. For small row counts, figure height uses a tighter internal heuristic to reduce excessive whitespace. Long footer text is wrapped and capped to 3 lines with ellipsis for overflow protection. Predictor labels are truncated (with warning) when they exceed layout-specific caps:

1. show_general_stats=True + two outcomes: 21 chars
2. show_general_stats=True + one outcome: 24 chars
3. show_general_stats=False + two outcomes: 26 chars
4. show_general_stats=False + one outcome: 25 chars When general stats are shown, large n/N values are compacted (e.g., 78,6k) to preserve column readability. Compaction activates only when counts reach >= 10.000 and uses a shared unit across both n and N (k, M, B, T) for consistent within-row formatting. Very large values beyond display range are capped as >999T with a warning. Rows are fully grayed only when all displayed outcomes are missing; if at least one outcome is valid, only the missing outcome triplet (effect, 95% CI, p) is blanked and gray-marked.

Exponentiation Safety

• Use exponentiate=None (default) for model/link-based automatic handling.
• Use exponentiate=False if your input is already on effect scale (e.g., OR/Ratio, not log-coefficients).
• Use exponentiate=True only when input is definitely on log scale and needs transformation.
• Read warnings: they include auto-exponentiation context and column mapping (effect column + CI_low/CI_high combined into 95% CI).

normalize_model_output()

clean_df, config = fpx.normalize_model_output(
    df, model_type="binom", link=None, exponentiate=None
)

Standardizes columns, applies exponentiation policy, and returns axis metadata. config includes exponentiated and renamed_columns for transparency.

Examples

Category grouping

df["category"] = ["Demographics", "Demographics", "Clinical", "Clinical"]

fig, axes = fpx.forest_plot(df, model_type="binom")

Dual outcomes

# DataFrame with two outcomes per predictor
fig, axes = fpx.forest_plot(
    df_two_outcomes,
    model_type="binom",
    outcomes=["Mortality", "Readmission"],
    legend_labels=["30-day mortality", "90-day readmission"],
)

Custom marker colors

fig, axes = fpx.forest_plot(
    df_two_outcomes,
    model_type="binom",
    outcomes=["Mortality", "Readmission"],
    point_colors=["#2C5F8A", "#D4763A"],
)

Linear model

fig, axes = fpx.forest_plot(df_linear, model_type="linear")

Save to file

fig, axes = fpx.forest_plot(df, model_type="binom", save="forest_plot.png")

Programmatic use (no display)

fig, axes = fpx.forest_plot(df, model_type="binom", show=False)
# Further customization...
fig.suptitle("My Forest Plot", fontsize=16)
fig.savefig("custom_plot.pdf", dpi=300)

In notebooks, show=False prevents internal plt.show(), but Jupyter may still auto-render the returned figure object. Use plt.close(fig) to suppress display.

Testing

The test suite lives in tests/ and covers all internal modules with no image comparisons — structural and behavioral assertions only.

Install dev dependencies first (see Installation), then:

pytest

Test files

File	Module under test	Tests
tests/test_normalization.py	_normalize.py	11
tests/test_layout.py	_layout.py	33
tests/test_axes_config.py	_axes_config.py	65
tests/test_plot_smoke.py	plot.py	4

Coverage summary

test_layout.py — build_row_layout()

• Flat layout (no category column): sequential y-positions, correct row count, all is_cat=False, "Uncategorized" labels, predictor order preserved, required columns present
• NaN predictor rows dropped; empty DataFrame raises ValueError
• Categorized layout: category header rows inserted, total = categories + predictors (parametrized), correct is_cat flags and per-predictor category labels, all-NaN category falls back to flat
• Dual-outcome DataFrames: unique() deduplication keeps one row per predictor regardless of outcome count

test_axes_config.py — configure_forest_axis() and helpers

• _nice_linear_step: 8 parametrized input→output pairs, zero, negative, tiny positive values
• _decimals_from_ticks: empty/single-tick → 2, step-inferred decimals (0/1/2), max_decimals cap
• Reference line: axvline placed at correct x for logit (1.0), log (1.0), identity (0.0); #910C07 color; dashed style; threshold override
• X-scale: "log" for logit/log links, "linear" for identity; empty data and thresholds=None do not crash
• X-label: correct label per link ("Odds Ratio" / "Ratio" / "β (coefficient)"), threshold override, font size propagated
• Y-ticks cleared; y-limits applied from thresholds["y_limits"]
• Spine visibility: top/right/left hidden, bottom visible
• X-limits contain full data range for log and linear axes; negative reference raises ValueError; span=0 edge case handled
• End-to-end parametrized across all four model types: binom, gamma, linear, ordinal
• show_general_stats=True/False both produce consistent output (documents no-op behaviour on axis)
• Tick count heuristic: num_ticks in {3, 5, 7} for log and linear axes
• tick_style="power10" uses readable rounded log10 exponents; single vs dual outcome lo_all/hi_all arrays both handled

Scope

forestplotx v1.0 is intentionally focused. It produces static, publication-quality forest plots for common regression model types.

Not included: interactive plots, Cox/Poisson models, theming engine, or GUI.

Versioning

forestplotx follows semantic versioning (SemVer).

• MAJOR – breaking API changes
• MINOR – backward-compatible feature additions
• PATCH – bug fixes and internal improvements

Current version: 1.0.1

License

MIT