🧠 Suicide Prediction System: A Socioeconomic Analysis

This Suicide Prediction System leverages data science 🧮 to analyze various environmental and socioeconomic factors such as GDP, weather conditions, happiness index, suicide rates, and population. By examining these indicators, the system identifies trends and patterns in suicide rates and develops predictive models to assess suicide risks. These insights can help organizations and mental health professionals take proactive measures to prevent suicides and improve public health outcomes. 🚑

---

🌐 Live Website

👉 Check out the live demo: Suicide Prediction Website

---

📊 Data Collection, Preparation, Cleaning & Visualizations

📥 Data Sources

• GDP 💰
• Weather (Temperature) 🌡️
• Happiness Index 😊
• Suicide Rates ⚰️
• Population 👥

🧹 Data Preparation

• ✅ Web scraping & API integration.
• ✅ Missing value imputation.
• ✅ Outlier detection & removal.
• ✅ Dataset merging & standardization.

🖼️ Sample Data

Before Cleaning

• GDP Data

Country	1999	2000	2001	2002
Country A	162.64	160.88	161.32	164.12

• Weather Data

Country	1999	2000	2001	2002
Country A	13	15	14	14.5

• Happiness Index

Country	Overall Rank	Life Evaluation
Country A	4	1.7666

• Suicide Rates

Country	Male	Female	Both
Country A	4.4	2.4	3.2

• Population

Country	1988	1989	1990
Country A	720859132	720907282	720889272

After Cleaning

Country	GDP	Happiness	Male Suicide	Female Suicide	Both Suicide	Population	Temperature	Year
Country A	165	1.766	4.4	2.4	3.2	720859132	15	1988

---

📈 Insights from Visualizations

• 📊 Bar Charts: Suicide rates across countries.
• 🔬 Scatter Plots: Relationship between population and suicide rates.
• 🌡️ Heatmaps: Correlation matrix of socioeconomic factors.

---

🤖 Model Implementation

Models Used

• 🎯 XGBoost
• 🌲 Random Forest
• 📍 K-Nearest Neighbors (KNN)
• 🚀 LightGBM
• ⚙️ Support Vector Machines (SVM)

📌 Conclusion

• ✅ KNN achieved the highest accuracy of 97%.
• ✅ Random Forest achieved 93.9%.
• ✅ Key predictive factors: GDP, Population Size, Happiness Index.
• ✅ Medium-risk classification remains challenging and needs further optimization.
• ✅ Future work includes feature engineering, additional data sources, and real-time prediction models.

---

🚀 Future Scope

• Improve model accuracy on medium-risk group.
• Add additional behavioral and psychological data.
• Build real-time prediction dashboard for public health monitoring.

---

📞 Contact

Atharv Kadam
📧 atharva895@gmail.com

---