DEV Community

Cover image for Native Machine Learning in Dart: `ml_knn` and `ml_logistic_regression` are Live on pub.dev!
Mehmet Çelik
Mehmet Çelik

Posted on

Native Machine Learning in Dart: `ml_knn` and `ml_logistic_regression` are Live on pub.dev!

#dart #flutter #opensource #pubdev

🧠 Introduction

The Dart ecosystem is growing fast — especially with Flutter. But when it comes to native machine learning (ML) capabilities inside Dart itself, the resources are almost nonexistent.

To address this gap, I’ve started developing a set of lightweight, native, and open-source machine learning packages in Dart. And now, the first two are officially published on pub.dev:

📦 What’s Available

1. ml_knn: K-Nearest Neighbors in Pure Dart

Use Cases:

  • Simple classification
  • Anomaly detection
  • Recommendation systems

Installation:

dependencies:
  ml_knn: ^1.0.0
Enter fullscreen mode Exit fullscreen mode

Usage:

final model = KNN(k: 3);
model.fit([[1.0, 2.0], [2.0, 3.0], [3.0, 4.0]], ['A', 'B', 'B']);
final prediction = model.predict([[2.5, 3.5]]);
print(prediction); // ['B']
Enter fullscreen mode Exit fullscreen mode

✅ Fully written in Dart

✅ Unit-tested and published on pub.dev

✅ Offline & mobile-friendly

2. ml_logistic_regression: Logistic Regression in Dart

Use Cases:

  • Binary classification
  • Probability-based predictions
  • Linear decision boundaries

Installation:

dependencies:
  ml_logistic_regression: ^1.0.0
Enter fullscreen mode Exit fullscreen mode

Example:

final model = LogisticRegression(
  learningRate: 0.1,
  iterations: 1000,
  regularization: 0.01,
);

model.fit([[0, 0], [1, 1]], [0, 1]);
final prediction = model.predict([[0.5, 0.5]]);
print(prediction); // [0] or [1]
Enter fullscreen mode Exit fullscreen mode

🧪 The model is validated with logical gate (AND) prediction tests

📈 All training is done natively in Dart – no Python, no API, no external libs.

🌱 Roadmap

This is just the beginning. Here's what I'm building next:

Package Status Description
ml_knn ✅ Live K-Nearest Neighbors (classification)
ml_logistic_regression ✅ Live Logistic regression
ml_fuzzy_matcher 🚧 In Progress AI-powered string similarity
ml_naive_bayes 🔜 Planned Naive Bayes Classifier
ml_linear_regression 🔜 Planned Linear regression
ml_kmeans 🔜 Planned Unsupervised clustering
ml_fin_scorer 🔜 Planned AI-powered financial scoring

🚀 Eventually, all will be part of a complete framework:

ml_flutter_basics

🔍 Why Native ML in Dart?

  • No need to host an API or use Python backends
  • Works offline, ideal for mobile apps
  • Seamless experience inside the Flutter ecosystem
  • Lightweight and fast to prototype in Dart

🤝 Contribute or Collaborate

I’m actively maintaining these projects and open to contributions!

✨ If you like the project, please give it a star!

💬 Issues, ideas, PRs — all are welcome.

📢 Final Thoughts

Machine learning in Dart should be accessible, lightweight, and usable in real-world apps — without always needing to call an external service or switch languages.

I’ll keep building these tools and sharing them publicly.

Thanks for reading!

Let’s make Dart smarter — together.

🙋‍♂️ About Me

I'm a Flutter & AI developer building native ML tools in Dart.

Also working on productivity apps like MergeNius, GreenPact, and more.

Building in public. Open to collaboration.

Let’s connect!

GitHub

Top comments (0)