MattTools

A Python toolkit for machine learning and bioinformatics: statistical analysis, model evaluation, and visualization.

Installation

# pip
pip install git+https://github.com/mattmuller0/MattTools.git

# uv
uv pip install git+https://github.com/mattmuller0/MattTools.git

For development:

git clone https://github.com/mattmuller0/MattTools.git
cd MattTools
pip install -e ".[dev]"

Quick Start

import matttools as mt
import numpy as np
from sklearn.ensemble import RandomForestClassifier
from sklearn.linear_model import LogisticRegression

# Set random seed across numpy, random, and ML frameworks
mt.set_random_seed(42)

# Calculate confidence intervals
data = np.random.normal(100, 15, 100)
mean, ci = mt.mean_confidence_interval(data, confidence=0.95)
print(f"Mean: {mean:.2f}, 95% CI: {ci}")

# Bootstrap resampling (sklearn-compatible splitter)
bootstrap = mt.Bootstrap(n_bootstrap=100, rng_seed=42)
for train_idx, test_idx in bootstrap.split(X, y):
    model.fit(X[train_idx], y[train_idx])

# Cross-validation with multiple models
models = {'rf': RandomForestClassifier(), 'lr': LogisticRegression()}
results = mt.modeling.cross_val_models(models, X, y, cv=5)

# Visualization with confidence intervals
mt.plotting.plot_roc_curve_ci(model, X_test, y_test, n_bootstraps=1000)
mt.plotting.plot_reduction(X, y, method='pca')  # PCA, UMAP, t-SNE

# Utility functions
result, elapsed = mt.utils.stopwatch(expensive_function, *args)
print(f"Completed in {elapsed:.2f}s")

API Reference

stats

Statistical functions with bootstrap and confidence interval support.

Function	Description
mean_confidence_interval(data, confidence)	Calculate mean with confidence interval
bootstrap_auc_confidence(y_true, y_score, n_bootstraps)	Bootstrap AUC with CI
Bootstrap(n_bootstrap, rng_seed)	sklearn-compatible bootstrap splitter
odds_ratio(table)	Calculate odds ratio from 2x2 contingency table

modeling

Model training, cross-validation, and evaluation utilities.

Function	Description
train_models(models, X_train, y_train)	Train multiple models
cross_val_models(models, X, y, cv)	Cross-validate multiple models
test_models(models, X_test, y_test)	Evaluate trained models on test set

plotting

Visualization tools for model evaluation and dimensionality reduction.

Function	Description
plot_reduction(X, y, method)	PCA/UMAP/t-SNE scatter plots
plot_scree(pca)	Scree plot for PCA variance
plot_roc_curve(y_true, score)	ROC curve
plot_roc_curve_ci(model, X, y, n_bootstraps)	ROC curve with bootstrap CI
plot_prc_curve(model, X, y)	Precision-recall curve
plot_pr_curve_ci(model, X, y, n_bootstraps)	PR curve with bootstrap CI
plot_confusion_matrix(y_true, score)	Confusion matrix heatmap
plot_confusion_matrices(models, X, y)	Multiple confusion matrices
plot_roc_curves(models, X, y)	Compare ROC curves
plot_prc_curves(models, X, y)	Compare PR curves
plot_model_results(results)	Visualize cross-validation results
plot_decision_boundaries(model, X, y)	Decision boundary plot
plot_cross_validation_auroc(model, X, y, cv)	CV AUROC with variance

utils

General utility functions.

Function	Description
set_random_seed(seed)	Set seed for numpy, random, torch, tensorflow
hide_warnings()	Suppress common ML warnings
get_memory_usage()	Get current memory usage in MB
print_memory_usage(label)	Print labeled memory usage
stopwatch(func, *args, **kwargs)	Time function execution

Requirements

Core: numpy, pandas, matplotlib, seaborn, scikit-learn, scipy, statsmodels

Optional: tensorflow, torch, umap-learn, plotly, bokeh, biopython, pysam, pyBigWig

Testing

pytest                    # Run all tests
pytest -v                 # Verbose output
pytest --cov=matttools    # With coverage

License

MIT License - Matthew Muller (matt.alex.muller@gmail.com)