🕵️ Crime Pattern Analysis and Predictive Modeling for Urban Safety

This project focuses on analyzing real-world crime data from Los Angeles and applying predictive modeling to identify high-risk crime zones and patterns. Using machine learning (Random Forest Classifier), data cleaning, and geospatial visualization, the goal is to assist urban law enforcement with insights for proactive crime prevention.

---

📊 Dataset

Source: Crime Data from 2020 to Present

Category	Description
Records	Crime reports from LAPD (Los Angeles Police Department)
Time Period	2020 to Present
Key Features	Date, Time, Location, Area, Weapon, Crime Type, Victim Demographics
Total Columns	25+ fields including categorical, numerical, and location-based features

---

🧹 Data Processing Pipeline

graph TD;
    A[Load Raw Crime Data] --> B[Data Cleaning];
    B --> C[Data Preprocessing];
    C --> D[EDA];
    D --> E[Split into Train/Test Sets];
    E --> F[Model Training: Random Forest];
    F --> G[Predict Probabilities];
    G --> H[Visualization & Evaluation];

Loading

---

🔍 Exploratory Data Analysis (EDA)

Victim Age Distribution

• Highest crime victim count in Adults (>60,000)
• Followed by Infants/Toddlers, then Seniors

Crime by Ethnicity

• Top 3 victim groups: Hispanic > Black > Other
• Over 50,000 cases for Hispanic victims

Geographic Crime Heat

• Central area is the highest crime zone (~14k cases)
• Followed by Southwest, Pacific, and North Hollywood

---

🤖 Model: Random Forest Classifier

• Input: Cleaned and transformed dataset
• Output: Crime prediction probabilities by category and region
• Evaluation: Accuracy, probability calibration, visualization

Notable Insights:

• Violent crimes against females peak in 77th Street division
• Sex crimes against males are highest in Foothill
• Central sees overall high male victimization rates

---

🔮 Future Enhancements

• Add real-time data integration (API-based)
• Include external factors: weather, events, socioeconomic data
• Deploy as a web-based crime risk dashboard
• Use geospatial clustering to detect micro-hotspots

---

📁 Files Included

File Name	Description
us_crime_data_final.ipynb	Jupyter Notebook with entire workflow
LAPD-Dataset-Link.pdf	Data schema and field definitions

---

📽️ Video Demo

👉 Watch the Project Video

Built for urban safety, backed by data.