simple-sklearn 0.1.0

Pythonic machine learning algorithms built for educational transparency and scikit-learn compatibility

simple-sklearn

simple-sklearn is a Python package designed to provide clear, readable, and highly pythonic implementations of fundamental machine learning algorithms. Abstracting away from complex low-level optimizations, this library focuses on clarity and educational value using high-level libraries like numpy, pandas, and scipy.

Every model is designed to integrate seamlessly with scikit-learn's estimator API, inheriting from sklearn.base.BaseEstimator and the appropriate mixins (ClassifierMixin or ClusterMixin). This allows you to plug these simple implementations directly into scikit-learn pipelines and cross-validation workflows.

WARNING: Not for Production Use. This library is built for educational purposes and algorithmic transparency. The implementations prioritize readability and simplicity over execution speed and memory optimization.

---

Available Models

Classification:

• OneRClassifier: 1R (One Rule) classification.
• NaiveBayesClassifier: Categorical Naive Bayes classification.
• KNeighborsClassifier: K-Nearest Neighbors classification.
• DecisionTreeClassifier: Decision Tree classification using the ID3 algorithm.

Clustering:

• KMeans: K-Means clustering.
• KMedoids: K-Medoids clustering.
• DBSCAN: Density-Based Spatial Clustering of Applications with Noise.
• AgglomerativeClustering: Hierarchical agglomerative clustering.

---

Installation

Requirements:

• Python 3.10+

Install directly from PyPI using pip:

pip install simple-sklearn

Core Dependencies:

• scikit-learn (>= 1.6.1)
• numpy (>= 1.26)
• pandas (>= 2.2.3)
• scipy (>= 1.13.1)
• typing-extensions (>=4.1.0, <5.0)

---

Quick Start

Because simple-sklearn strictly implements the scikit-learn API, you can fit and predict models exactly as you would with scikit-learn.

Classification Example

import numpy as np
from simple_sklearn.classification import NaiveBayesClassifier

# Categorical data
X = np.array([[0, 0], [0, 1], [1, 0], [2, 2], [2, 3]])
y = np.array([0, 0, 0, 1, 1])

# Initialize and fit the model
clf = NaiveBayesClassifier()
clf.fit(X, y)

# Predict on new data
X_new = np.array([[2, 2], [1, 4]])
predictions = clf.predict(X_new)  # Handles unseen category '4' gracefully
print(f"Predictions: {predictions}")  # Output: [1 0]

Clustering Example

import numpy as np
from simple_sklearn.clustering import KMeans

# Continuous data
X = np.array([[0.1, 0.1], [0.2, 0.1], [10.1, 10.1], [10.2, 10.1]])

# Initialize and fit the model
clusterer = KMeans(n_clusters=2, max_iter=10, random_state=42)
clusterer.fit(X)

print(f"Cluster labels: {clusterer.labels_}")  # Output: [0 0 1 1]
print(f"Cluster centers: \n{clusterer.cluster_centers_}")  # Output: [[0.15, 0.1], [10.15, 10.1]]

---

Documentation

Detailed API references, algorithm details, and Jupyter Notebook usage examples are available on the library website:

Artem259.github.io/simple-sklearn

---

Development & Contributing

Pull requests are welcome! If you'd like to contribute, please read the Contributing Guide.

---

Roadmap

Future development focuses on:

1. Estimator Enhancements and Performance Optimization:
   • Implementing a custom KDTree for faster nearest-neighbor searches.
   • Refactoring AgglomerativeClustering with Priority Queue (Min-Heap) to reduce time complexity during cluster merging.
   • Adding other estimator features (e.g., standard hyperparameters like max_depth for DecisionTreeClassifier).
2. Documentation Upgrades:
   • Configuring documentation versioning via the mike tool.

---

License

This project is licensed under the MIT License.