End-to-End Airbnb Machine Learning Pipeline

This repository contains a complete, end-to-end machine learning pipeline for training and running predictions on Airbnb listing data.
The project is designed to reflect production-grade ML practices, where each stage of the pipeline is clearly separated, reproducible, and driven by explicit artifacts.

The pipeline starts from raw CSV files and ends with model predictions and summarized results. Business Goal: The machine learning model is specifically designed to predict the ratings of an Airbnb listing based on its features.

---

What This Repository Does (High-Level)

The workflow implemented in this repository follows this sequence:

1. Take raw Airbnb listing data from multiple cities
2. Clean and merge the datasets into a single processed dataset
3. Train a machine learning model using a robust preprocessing + modeling pipeline
4. Save the trained model as a reusable artifact
5. Use the trained model to generate predictions on data
6. Summarize prediction outputs into final results

Each step is implemented as a standalone Python script that can be run independently.

---

Repository Structure

.
├── data/
│   ├── raw/
│   │   ├── listingsLA.csv           # Raw Los Angeles Airbnb data
│   │   └── listingsNYC.csv          # Raw New York City Airbnb data
│   │
│   └── processed/
│       ├── listings_combined_clean.csv
│       └── cleaning_summary.json
│
├── models/
│   └── model.joblib                 # Trained ML pipeline
│
├── results/
│   ├── metrics.json                 # Training evaluation metrics
│   ├── predictions.csv              # Model predictions
│   └── prediction_summary.json      # Aggregated prediction results
│
├── preprocess.py                    # Data preprocessing script
├── train.py                         # Model training script
├── predict.py                       # Batch prediction script
├── results.py                       # Prediction aggregation script
├── requirements.txt
└── README.md

---

Installation & Setup

1. Clone the repository:

   git clone https://github.com/TalSomech/capstone_22.git
   cd capstone_22

2. Install requirements:

   pip install -r requirements.txt

Usage Instructions

To execute the pipeline from start to finish, run the scripts in the following order:

# 1. Clean and merge the raw data
python preprocess.py

# 2. Train the model and save the artifact
python train.py

# 3. Generate predictions using the saved model
python predict.py

# 4. Aggregate results and output summaries
python results.py