DeRent5/ai-service

A repository for Morocco Princing API model Microservice using FastAPI , alongside with 3 other AI Features # Morocco Airbnb Dynamic Pricing API AI-powered dynamic pricing microservice for Airbnb listings across Moroccan cities. Predicts optimal nightly prices using a Random Forest model trained on 1,656+ real listings. ## Overview This API delivers real-time price predictions for short-term rental properties in Morocco's major cities. Built with **FastAPI** and powered by a **Random Forest** machine learning model achieving **84.59 MAD mean absolute error** (~$8.46 USD). ### Key Features - **Fast Predictions**: Single prediction ` - Specific build - `medgm/morocco-pricing-api: ` - Git commit version --- ## Project Structure ``` pricing-model-api/ ├── data/ │ ├── raw/ # Raw JSON listings │ ├── processed/ # Processed datasets │ └── used_or_will_be_used/ │ └── all_listings_clean.csv # Clean dataset (1,656 listings) ├── deployment/ │ ├── Dockerfile # Production Docker image │ ├── Jenkinsfile # CI/CD pipeline │ ├── app.py # FastAPI application │ ├── docker-compose.yml # Docker Compose config │ ├── requirements.txt # Python dependencies │ └── test_api.py # API tests ├── models/ │ ├── production/ │ │ └── random_forest_tuned.pkl.gz # Trained model (~94.55 MB compressed) │ ├── model_comparison.csv # Training metrics summary │ └── rf_feature_importance.csv # Random Forest feature importances ├── notebooks/ │ ├── airbnb_pricing_pipeline.ipynb # ETL pipeline │ └── pricing_model_training.ipynb # Model training └── README.md # This file ``` --- ## Technology Stack ### Backend - **Framework**: FastAPI 0.115.6 - **Server**: Uvicorn (ASGI) - **Validation**: Pydantic v2 ### Machine Learning - **Algorithm**: RandomForestRegressor - **Library**: scikit-learn 1.7.2 - **Model Size**: 94.55 MB (compressed) - **Feature Space**: …