Understanding Public Safety Trends in Calgary: A Data Mining Perspective
- 1. Understanding Public Safety Trends in Calgary: A Data Mining Perspective - Manjunatha Inti Unveiling Crime Patterns and Trends Using Clustering, Predictive Models, and Geospatial Analysis
- 2. Introduction Why Public Safety Matters • Challenges of urbanization: crime, tra ff ic, and disorder. • Importance of data-driven insights for governance. • Objective: Leverage data mining to uncover safety patterns in Calgary.
- 3. What Does the Study Do? • Identify factors in f luencing public safety. • Uncover spatial and temporal patterns in crime, tra ff ic, and disorder. • Develop predictive models for resource allocation. • Figure: Fig. 1 (Crime Categories) to show dominant safety concerns.
- 4. Data Sources and Processing • Crime data, tra ff ic incidents, census data, infrastructure data, pet registrations. • Methods: Cleaning, geocoding, and feature engineering. • Figures: Fig. 2 (Tra ff ic Incidents) and a summary work f low diagram. Multi-Source Data Integration
- 5. Methodology • Correlation Analysis: Relationships between demographics and safety metrics. Figure: Fig. 3 (Correlation Heatmap for Census Data). • Clustering Techniques: K-Means, DBSCAN, and CLARANS. Figure: Fig. 3 (K-Means Clustering Results). • Predictive Modeling: Regression models for resource allocation. How It Works
- 7. Key Findings • Hotspot Communities: Beltline and Downtown Core report the highest number of crimes and disorders, with Forest Lawn being another critical area. These areas are heavily impacted due to population density and urban activities. • Fig. 4: Top 10 Communities with the Highest Total Number of Crimes (2018– 2023). • Fig. 5: Top 10 Communities with the Highest Total Number of Disorders (2018–2023). Geospatial Insights:
- 8. Key Findings • Seasonal Trends: Crimes peak during warmer months (May to August) due to increased outdoor activities and gatherings. Tra ff ic incidents, however, rise during winter months (November to December) because of poor weather conditions and driving challenges. • Fig. 6: Crime Trends Over Time by Category. • Fig. 7: Monthly Average Crimes (2018– 2023). Temporal Patterns
- 9. Key Findings • Regression models indicate that population density, apartment dwellings, and educational disengagement are the strongest predictors of crime and disorder. Communities with low school support systems exhibit higher disorder rates. • Fig. 8: Correlation Heatmap (Lights and Trees). Modeling Insights
- 10. Applications Practical Applications • Targeted interventions for high-risk communities. • Optimized resource allocation for policing and infrastructure. • Scalable framework for other cities. • Visual: Infographic or steps based on clustering and modeling results.
- 11. Strengths and Challenges Balancing Insights and Limitations • Strengths: 1. Comprehensive multi-source data integration. 2. Actionable insights for urban governance. • Challenges: 1. Data completeness and clustering overlaps. 2. Short-term analysis restricting long-term trends. • Figure: Tabular comparison of strengths and challenges
- 12. Strengths Challenges Integration of diverse datasets (crime, traf fi c, census, infrastructure, pets). Limited completeness in data (e.g., missing or derived geographic coordinates). Application of advanced algorithms (K-Means, DBSCAN, regression). Overlapping clusters with low silhouette scores. Identi fi cation of geospatial and temporal patterns in safety metrics. Short-term analysis window restricts long-term trends. Actionable insights for resource allocation and urban planning. Limited representation of socio-economic and cultural factors. Scalable methodology applicable to other cities. Complex data relationships require more re fi ned hyperparameters.
- 13. Limitations • Data Quality and Completeness: Some datasets had missing or derived geographic coordinates, which may have impacted the accuracy of clustering and analysis. • Short-Term Analysis Window: The study is based on data spanning a limited timeframe, restricting its ability to capture long- term trends and evolving patterns. • Clustering Challenges: Overlapping clusters and low silhouette scores indicate potential limitations in data separability or the need for re f ined hyperparameters. • Limited Socioeconomic Context: The study primarily focuses on quantitative metrics, with minimal representation of cultural or socioeconomic factors in f luencing public safety.
- 14. Conclusion Summary and Takeaways • Data mining is a powerful tool for public safety analysis. • Calgary’s insights pave the way for smarter cities. • Addressing limitations can enhance urban planning strategies.