Ridge Regression
Gradient Boost
Linear Regression
Key Insights
Our analysis revealed crucial patterns that optimize EV charging infrastructure deployment
Model Performance 87% Accuracy
XGBoost consistently outperforms other models in both classification and regression tasks with higher accuracy and lower error rates. The ensemble-based methods (Random Forest and XGBoost) demonstrate superior performance over single-model approaches.
Location Insights 35% Coverage Increase
Areas with high road density and population concentration show the strongest correlation with optimal charging station placement. Commercial zones located near residential areas with high EV ownership demonstrate the highest predicted utilization rates.
Key Findings Western Suburbs
The spatial distribution analysis reveals that Mumbai's western suburbs have the highest potential for new charging stations, particularly in areas with developing smart grid infrastructure. Additionally, existing station density is inversely correlated with predicted success of new stations beyond a certain threshold.
Usage Patterns 9am-5pm Peak
Temporal analysis shows peak charging demand occurs during working hours (9am-5pm), with secondary peaks in early evening. Weekend usage follows different patterns with more distributed charging throughout the day. These insights can inform dynamic pricing strategies and capacity planning.
Conclusion
The machine learning models demonstrate that optimizing EV charging station placement requires a multi-faceted approach that considers infrastructure, demographics, and existing usage patterns. Our XGBoost model achieved the highest accuracy (87%) in classification tasks and lowest error rates in regression tasks, making it the recommended approach for deployment.
These insights have been translated into an interactive map highlighting the top 20 recommended locations for new EV charging stations in Mumbai, with an estimated increase in coverage of approximately 35% while minimizing overlap with existing infrastructure.