🧱 LEGO Set Identifier with K-Nearest Neighbors 🧩

This project uses a K-Nearest Neighbors (KNN) model to identify potential LEGO set IDs based on a list of LEGO parts you provide. With a trained model and a parts list, you can find matching LEGO sets from your collection! Perfect for enthusiasts, collectors, and developers. 🎉

✨ Features

• 🔍 LEGO Set Matching: Input a dictionary of parts and quantities, and the KNN model will return potential matching set IDs.
• 🧠 Cosine Similarity Matching: Utilizes cosine similarity to find the closest matches.
• 💾 Save and Load Models: Trained models and scalers are saved for easy re-use.

🚀 Getting Started

📋 Prerequisites

• Python 3.12
• pandas, scikit-learn, numpy, joblib (Install with pip install -r requirements.txt)

📥 Installation

Setup with Pixi

Install Pixi if not already installed:

Initialize your environment:

pixi install

This will automatically install the required dependencies listed in pyproject.toml

🛠️ Usage

Ensure dataframe.pkl exists, containing LEGO set data with part quantities. Also, the files knn_model.pkl and scaler.pkl are both needed. They can be generated by running

tar -xvzf df.pkl.tgz datafram.pkl
tar -xvzf knn.pkl.tgz knn_model.pkl
tar -xvzf scaler.pkl.tgz scaler.pkl

Run the Script:

Train the model and predict sets for a parts list:

pixi run python main_model.py

Make Predictions:

Example dictionary format for parts: python Copy code parts_list = { '6141': 2, '3001': 4, '3003': 1 } Expected Output:

Closest set IDs for the given parts list: json Copy code { "set_ids": ["1234-1", "5678-1"] } 📤 Example Request Using curl:

curl -X POST http://localhost:5000/identify -H "Content-Type: application/json" -d '{"parts": {"6141": 2, "3001": 4}}'

📂 Files in Project

• dataframe.pkl: Pickle file with LEGO set data.
• knn_model.pkl: Saved KNN model file.
• scaler.pkl: Saved scaler for normalizing parts quantities. 📜 License This project is open-source. Feel free to use, modify, and share! 🎉