🚆 RMS_Models

---

📌 Overview

The RMS_Models repository provides machine learning models and services for railway operations analytics, focusing on:

• ⏱️ Dwell Time Prediction – estimating how long a train spends at a station.
• 🕒 ETA/ETD Modeling (ETADS) – predicting expected arrival and departure times.
• 🧩 Ensemble Learning – combining multiple models to improve accuracy.
• 🔄 Data preprocessing pipelines – cleaning and preparing raw railway datasets.

This project is built for both research experiments (Jupyter notebooks) and production deployment (Python services + model artifacts).

---

✨ Features

• Machine learning models (Random Forest, Ensemble) for dwell time prediction.
• ETA/ETD model for real-time train arrival/departure estimation.
• REST API services (Flask or FastAPI) for model deployment.
• Modular code structure for easy experimentation.
• Preprocessed datasets (Live_Trains_Fixed_Clean.csv) for reliable training.
• Training logs for reproducibility.

---

📂 Repository Structure

RMS_models/ │── .gitattributes │── README.md │── pycache/ │ │── models/ │ ├── ensemble_dwell_time_model.pkl │ ├── ensemble_dwell_time_model(Self).pkl │ ├── rf_model.pkl │ ├── ETADS.pkl │ │── datasets/ │ ├── Block_Relation.csv │ ├── Blocks.csv │ ├── Stations.csv │ ├── Trains.csv │ ├── Live_Trains.csv │ ├── Live_Trains_Fixed_Clean.csv │ │── notebooks/ │ ├── DwellTime(2).ipynb │ ├── DwellTime(Draft).ipynb │ ├── ETA.ipynb │ │── scripts/ │ ├── dwellTime.py │ ├── dwell_time_ml_service.py │ ├── ml_service.py │ │── logs/ │ ├── train_model.log

yaml Copy code

---

⚙️ Models

🔹 Dwell Time Model

• Goal: Predict how long a train will stay at a station.
• Techniques: Random Forest + Ensemble.
• Inputs: Train ID, Station, Time, Block data.
• Output: Predicted dwell time (in minutes).

Artifacts:

• rf_model.pkl → Random Forest model.
• ensemble_dwell_time_model.pkl → Final ensemble model.

---

🔹 ETADS Model

• Goal: Predict Expected Time of Arrival (ETA) & Departure (ETD).
• Inputs: Station sequences, block relations, live train status.
• Output: ETA and ETD predictions.

Artifact:

• ETADS.pkl

---

📊 Datasets

• Stations.csv → Station metadata.
• Blocks.csv, Block_Relation.csv → Railway block definitions and connectivity.
• Trains.csv → Train metadata.
• Live_Trains.csv → Raw live data.
• Live_Trains_Fixed_Clean.csv → Preprocessed dataset used in modeling.

---

🚀 Installation & Setup

1️⃣ Clone Repository

git clone https://github.com/<your-username>/RMS_models.git
cd RMS_models
2️⃣ Install Requirements
bash
Copy code
pip install -r requirements.txt
3️⃣ Train Models
bash
Copy code
python scripts/dwellTime.py
4️⃣ Run Dwell Time API Service
bash
Copy code
python scripts/dwell_time_ml_service.py
🔧 API Usage
📌 Endpoint: /predict_dwell_time
Request (JSON):

json
Copy code
{
  "train_id": 101,
  "station_id": "NDLS",
  "arrival_time": "2025-09-25T10:30:00"
}
Response (JSON):

json
Copy code
{
  "predicted_dwell_time": 7.5
}
Example with curl
bash
Copy code
curl -X POST http://127.0.0.1:5000/predict_dwell_time \
     -H "Content-Type: application/json" \
     -d '{"train_id":101,"station_id":"NDLS","arrival_time":"2025-09-25T10:30:00"}'
📈 Experiments
DwellTime(Draft).ipynb → Feature engineering + baseline experiments.

DwellTime(2).ipynb → Final tuned dwell time model.

ETA.ipynb → ETADS prototype (arrival/departure estimation).

📝 Logs
train_model.log contains:

Training progress.

Hyperparameters used.

Evaluation metrics (MAE, RMSE, R²).