Altifrax

(Airplane Accident Severity Prediction) 🚨

🧠 Overview

This project aims to predict the severity of airplane accidents using machine learning. The model analyzes various flight, safety, and environmental features to classify accidents into different severity categories, helping to assess risk and prevent future incidents.

---

⚙️ Workflow

1. Data Loading

   • The training dataset (trn.csv) and test dataset (tst.csv) are loaded using Pandas.

2. Data Preprocessing

   • Checks for missing values.
   • Encodes the target variable Severity using Label Encoding.
   • Drops non-essential columns such as Accident_ID and Adverse_Weather_Metric.

3. Exploratory Data Analysis (EDA)

   • Generates a Pearson correlation matrix to identify relationships between numerical features.
   • Visualizes correlations with a Seaborn heatmap to understand which variables most affect severity.

4. Feature Engineering

   • Selects highly correlated features for better model accuracy.
   • Scales continuous variables like Max_Elevation, Safety_Score, and Accident_Type_Code using MinMaxScaler.

5. Model Training

   • Splits the data (90% training, 10% testing).
   • Trains an XGBoost Classifier, a gradient-boosting algorithm optimized for structured data.

6. Prediction & Evaluation

   • Predicts severity levels for the test dataset.

   • Maps numeric predictions to descriptive categories:

     • 0 → Highly Fatal and Damaging
     • 1 → Minor Damage and Injuries
     • 2 → Significant Damage and Fatalities
     • 3 → Significant Damage and Serious Injuries

7. Output Generation

   • Saves predictions to a CSV file (Q130.csv) containing Accident_ID and corresponding severity predictions.

---

📦 Tech Stack

• Languages: Python
• Libraries: Pandas, NumPy, Scikit-learn, XGBoost, Seaborn, Matplotlib, Statsmodels
• Visualization: Correlation Heatmaps

---

📊 Results

The model successfully classifies airplane accidents into four severity categories based on the given features, demonstrating effective feature analysis and predictive modeling using XGBoost.

---

🧩 Future Enhancements

• Hyperparameter tuning for improved accuracy.
• Incorporating additional weather and aircraft-specific parameters.
• Deployment as a web-based prediction dashboard.