🚀 MultiModel Analysis

Train • Evaluate • Compare • Visualize Multiple Machine Learning Models in Minutes

A lightweight, robust Python library that automates model benchmarking, metric evaluation, and visualization for both Classification and Regression tasks.

Created & Maintained by Uditya Narayan Tiwari

📖 Table of Contents

• Overview
• Why MultiModel Analysis?
• Key Features
• Supported Models
• Installation
• Quick Start
  • Classification Example
  • Regression Example
• End-to-End Workflow
• API Reference
  • MultiModelClassifier
  • MultiModelRegressor
• Requirements
• License & Citation

📖 Overview

Selecting the optimal Machine Learning model requires training, tuning, and comparing multiple algorithms. Traditionally, this process demands writing repetitive boilerplate code for:

• Splitting datasets and feature scaling
• Fitting each estimator individually
• Extracting evaluation metrics (Accuracy, F1, MAE, R², etc.)
• Generating figures (Confusion Matrices, ROC Curves, Scatter plots)
• Ranking models to recommend the best performer

MultiModel Analysis automates this entire pipeline into a clean, intuitive, production-grade interface. With just a few lines of code, you can train 8 classifiers or 7 regressors, evaluate their performance, generate publication-ready plots, and receive an instant recommendation for the best model.

💡 Why MultiModel Analysis?

Instead of writing repetitive, error-prone code like this:

# Traditional Workflow (100+ lines of repetitive code)
model1.fit(X_train, y_train)
model2.fit(X_train, y_train)
pred1 = model1.predict(X_test)
pred2 = model2.predict(X_test)

acc1 = accuracy_score(y_test, pred1)
f1_1 = f1_score(y_test, pred1, average='weighted')
# ... Repeat for every model & plot manually

Simply write:

from multimodel_analysis import MultiModelClassifier

# Automated Benchmarking Workflow
classifier = MultiModelClassifier(X, y, scaled_data=True)
results = classifier.run_all_models()
classifier.get_summary(results)

✨ Key Features

• ⚡ Automated Multi-Model Benchmarking: Train up to 8 classifiers or 7 regressors simultaneously.
• 🎯 Multiclass & String Target Support: Built-in LabelEncoder handles string targets (e.g. 'Cat', 'Dog'), multi-class target variables, and binary targets smoothly.
• 📊 Multiclass ROC-AUC Calculation: Accurate One-vs-Rest ROC-AUC computation (multi_class='ovr') for multiclass classification without silent metric failures.
• 📏 DataFrame Feature Integrity: Preserves pandas DataFrame column names and indices during standard scaling.
• 🛡️ Fault-Tolerant Execution: Exception handling insulates model fitting so a single failing estimator will not crash the overall benchmark.
• 🖥️ Cross-Platform Safe: Unicode print fallbacks prevent charmap / terminal encoding errors on Windows, macOS, and Linux console environments.
• 📈 Publication-Ready Visualizations: Automatically generates styled Confusion Matrices with class names, ROC curves, Regression error scatter plots, and metric bar charts.

🤖 Supported Models

🎯 Classification Models (MultiModelClassifier)

1. Logistic Regression
2. Support Vector Machine (SVC)
3. K-Nearest Neighbors (KNN)
4. Decision Tree Classifier
5. Random Forest Classifier
6. Gaussian Naive Bayes
7. Gradient Boosting Classifier
8. AdaBoost Classifier

📈 Regression Models (MultiModelRegressor)

1. Linear Regression
2. Lasso Regression
3. Ridge Regression
4. Support Vector Regression (SVR)
5. Decision Tree Regressor
6. Random Forest Regressor
7. Gradient Boosting Regressor

(Note: MultiModelRegressior is retained as a backwards-compatible alias for legacy code).

📦 Installation

From PyPI (Recommended)

pip install multimodel-analysis

Upgrade to the latest version:

pip install --upgrade multimodel-analysis

From GitHub Source

pip install git+https://github.com/udityamerit/Multimodel-Analysis-Pacakge.git

🚀 Quick Start

Classification Workflow

import pandas as pd
from multimodel_analysis import MultiModelClassifier

# 1. Load your dataset
df = pd.read_csv("dataset.csv")
X = df.drop("target", axis=1)
y = df["target"]  # Can be numeric or strings like 'Class_A', 'Class_B', 'Class_C'

# 2. Initialize classifier with feature scaling & stratified train/test split
classifier = MultiModelClassifier(
    X=X, 
    y=y, 
    test_size=0.3, 
    scaled_data=True, 
    random_state=42
)

# 3. Train all classification models
results = classifier.run_all_models()

# 4. Show tabular summary and visualizations
classifier.show_tabular_report(results)
classifier.plot_confusion_matrices(results)
classifier.plot_roc_curves(results)
classifier.plot_comparison(results)

# Or run all reporting functions in one call:
# classifier.get_summary(results)

Regression Workflow

import pandas as pd
from multimodel_analysis import MultiModelRegressor

# 1. Load housing dataset
df = pd.read_csv("housing.csv")
X = df.drop("Price", axis=1)
y = df["Price"]

# 2. Initialize regressor
regressor = MultiModelRegressor(
    X=X, 
    y=y, 
    test_size=0.3, 
    scaled_data=True, 
    random_state=42
)

# 3. Train all regression models
results = regressor.run_all_models()

# 4. Display report table & plots
regressor.show_tabular_report(results)
regressor.plot_true_vs_predicted(results)
regressor.plot_comparison(results)

🏗 End-to-End Workflow

flowchart LR
    A["📂 Load Dataset (X, y)"] --> B["🧹 Target & Data Preprocessing"]
    B --> C["🏷 LabelEncoder (String & Multiclass)"]
    C --> D{"⚙️ Feature Scaling?"}
    D -->|Enabled| E["📏 StandardScaler (Preserves Columns)"]
    D -->|Disabled| F["➡️ Raw Features"]
    E --> G["✂️ Stratified Train / Test Split"]
    F --> G
    G --> H{"🎯 Machine Learning Task"}
    H -->|Classification| I["🤖 Train 8 Classifiers"]
    H -->|Regression| J["📈 Train 7 Regressors"]
    I --> K["📊 Compute Accuracy, F1, ROC-AUC (OVR)"]
    J --> L["📈 Compute MAE, MSE, RMSE, R² Score"]
    K --> M["📈 Generate Figures & Tabular Benchmark"]
    L --> M
    M --> N["🏆 Recommend Best Model"]

📚 API Reference

MultiModelClassifier

MultiModelClassifier(X, y, test_size=0.3, scaled_data=False, random_state=42, stratify=True)

Parameters:

• X: DataFrame or array-like of shape (n_samples, n_features) — Feature matrix.
• y: Series or array-like of shape (n_samples,) — Target labels (numeric or categorical strings).
• test_size: float, default=0.3 — Proportion of dataset for test split.
• scaled_data: bool, default=False — Fits and applies StandardScaler to features.
• random_state: int, default=42 — Random seed for reproducibility.
• stratify: bool, default=True — Enables stratified splitting for balanced class ratios.

Key Methods:

• .run_all_models(): Fits all classification algorithms and returns evaluated metric tuples.
• .show_tabular_report(models): Prints clean comparison table sorted by Accuracy and recommends the top model.
• .plot_confusion_matrices(models): Displays styled confusion matrix heatmaps with actual class labels.
• .plot_roc_curves(models): Plots combined ROC curves and AUC scores.
• .plot_comparison(models): Generates metric comparison bar plots (Accuracy, Precision, Recall, F1).
• .get_summary(models): Runs complete reporting and plotting pipeline.

MultiModelRegressor

MultiModelRegressor(X, y, test_size=0.3, scaled_data=False, random_state=42)

Parameters:

• X: DataFrame or array-like of shape (n_samples, n_features) — Feature matrix.
• y: Series or array-like of shape (n_samples,) — Continuous target variable.
• test_size: float, default=0.3 — Proportion of dataset for test split.
• scaled_data: bool, default=False — Fits and applies StandardScaler to features.
• random_state: int, default=42 — Random seed for reproducibility.

Key Methods:

• .run_all_models(): Fits all regressor algorithms and returns evaluation metric tuples.
• .show_tabular_report(models): Displays tabular report sorted by $R^2$ Score and recommends the top regressor.
• .plot_true_vs_predicted(models): Displays True vs Predicted value scatter plots with perfect prediction reference line.
• .plot_comparison(models): Displays $R^2$ score bar plot comparison across models.
• .get_summary(models): Runs complete reporting pipeline.

📚 Requirements

📜 License & Author

Distributed under the Apache 2.0 License. See LICENSE for more information.

Author & Maintainer:
Uditya Narayan Tiwari
📧 Email: tiwarimerit@gmail.com
🌐 GitHub: @udityamerit
📦 PyPI: multimodel-analysis