Crime Data Analytics & Prediction Project

A comprehensive time-series and geospatial analysis of crime data using R

Check it out at https://bog67.shinyapps.io/crime-analytics-dashboard/

📊 Project Overview

This project analyzes and forecasts crime trends using real-world police crime datasets (Chicago Crimes Data, Madison PD reports, etc.). It combines time-series forecasting, geospatial visualization, and interactive dashboards to provide actionable insights into crime patterns.

🎯 Key Features

• Data Import & Cleaning: Automated pipeline using tidyverse and janitor
• Time-Series Analysis: Trend detection and seasonal patterns with lubridate and tsibble
• Crime Forecasting: ARIMA and Prophet models for 6-month predictions
• Geospatial Heatmaps: Interactive crime hotspot maps using ggmap and leaflet
• Interactive Dashboard: Comprehensive Shiny dashboard for visualization

� Screenshots

Dashboard Overview

Time-Series Analysis

Geospatial Visualization

�📈 Key Outputs

✅ Crime hotspots by location
✅ Monthly/seasonal crime patterns
✅ Crime forecast for next 6 months
✅ Interactive crime dashboard

Skills Demonstrated

• Data cleaning and preprocessing
• R data wrangling & feature engineering
• Time-series forecasting (ARIMA, Prophet)
• Geospatial visualization
• Interactive dashboard development
• Shiny dashboard development
• Statistical modeling

📁 Project Structure

Crime report/
│
├── data/                  # Raw and cleaned datasets
│   ├── raw/              # Original crime data
│   └── processed/        # Cleaned data
│
├── scripts/              # R analysis scripts
│   ├── 01_data_import.R          # Data loading and cleaning
│   ├── 02_time_series_analysis.R # Temporal analysis
│   ├── 03_forecasting.R          # ARIMA & Prophet models
│   ├── 04_geospatial.R           # Maps and heatmaps
│   └── 00_install_packages.R    # Package installation
│
├── outputs/              # Generated visualizations
│   ├── plots/           # Time-series charts
│   └── maps/            # Geospatial visualizations
│
├── dashboard/           # Shiny application
│   ├── app.R           # Main dashboard
│   └── ui.R            # UI components (if separated)
│
└── README.md           # This file

📦 Required Packages

Install all required packages by running:

source("scripts/00_install_packages.R")

Core Packages:

• tidyverse - Data manipulation and visualization
• janitor - Data cleaning
• lubridate - Date/time handling
• tsibble - Time-series data structures
• forecast - ARIMA modeling
• prophet - Facebook's forecasting tool
• ggmap - Static maps
• leaflet - Interactive maps
• shiny - Interactive dashboard
• flexdashboard - Dashboard layout
• sf - Spatial data handling
• viridis - Color palettes

📊 Analysis Workflow

Step 1: Data Import & Cleaning

• Load crime datasets
• Clean column names
• Handle missing values
• Standardize date formats
• Filter relevant crime types

Step 2: Time-Series Analysis

• Aggregate crimes by time period
• Detect trends and seasonality
• Create time-series visualizations
• Identify peak crime periods

Step 3: Forecasting

• Prepare time-series data
• Build ARIMA model
• Build Prophet model
• Compare model performance
• Generate 6-month forecasts

Step 4: Geospatial Analysis

• Clean location data
• Create crime density maps
• Identify hotspots
• Generate interactive leaflet maps

Step 5: Dashboard

• Integrate all visualizations
• Add interactive filters
• Display key metrics
• Enable data exploration

📈 Sample Outputs

Time-Series Forecast

• Monthly crime trends
• Seasonal decomposition
• 6-month ARIMA predictions
• Prophet forecast with confidence intervals

Geospatial Maps

• Crime hotspot heatmaps
• Location-based clustering
• Interactive marker maps
• Density overlays

Dashboard Features

• Date range selector
• Crime type filter
• Location filter
• Download reports

🔍 Key Insights

After running the analysis, you'll discover:

1. Temporal Patterns: When crimes peak (time of day, day of week, season)
2. Geographic Hotspots: High-crime areas requiring attention
3. Crime Categories: Most common crime types
4. Trends: Whether crime is increasing or decreasing
5. Forecasts: Expected crime levels for the next 6 months

📝 Notes

• Data Privacy: Ensure compliance with data usage policies
• API Keys: Google Maps API key required for ggmap (or use OpenStreetMap)
• Performance: Large datasets may require data sampling for faster processing
• Updates: Crime data is typically updated monthly