With the increase in the rate of crimes, there is raising concern among the citizens and the authorities. Understanding crime patterns can help law enforcement agencies allocate resources efficiently and enhance public safety. This blog gives detailed information on how AI and data science can be used to analyze the data about crimes and create an interactive heatmap to visualize the high crime places. Using Python, FireDucks, GeoPandas, and visualization tools such as Folium and Plotly, we can take an unprocessed dataset containing crime and convert the unprocessed dataset into an informative crime heatmap.

PROBLEM STATEMENT:-

Crime data is often scattered across multiple sources, making it difficult to analyze the trends effectively. By developing an AI-powered crime heatmap, we can provide valuable insights for policymakers, researchers, and public safety organizations.

SOLUTION/ APPROACH:-

Step 1: Collecting Crime Data (India)

In India, crime data is reported by agencies like the National Crime Records Bureau (NCRB) and local police departments. Some reliable sources for data collection include

• NCRB Reports: https://ncrb.gov.in/

• Open Government Data (OGD) Platform India: https://data.gov.in/

• State Police Portals (Delhi, Maharashtra, Tamil Nadu, etc.)

For this project, we will use NCRB’s city-wise crime data, which includes:

• Crime type (Murder, Theft, Assault, Cybercrime, etc.)

• Location (State, City, Latitude, Longitude)

• Time (Year, Month)

Step 2: Data Preprocessing & Cleaning (India Crime Data)

We will preprocess the NCRB data to ensure consistency using FireDucks.

import fireducks as fd
# Load dataset (Modify path based on the source)

df = fd.read_csv("ncrb_crime_data.csv")

# Convert timestamp

df["Year"] = fd.to_datetime(df["Year"], format="%Y")

# Remove missing values

df = df.dropna(subset=["Latitude", "Longitude"])

# Display the cleaned dataset

df.head()

Step 3: Exploratory Data Analysis (EDA) - India Crime Patterns

Key Insights from Indian Crime Data:

• Crime hotspots by city (Delhi, Mumbai, Bangalore, etc.)

• Types of crimes increasing over time (e.g., Cybercrime, Domestic Violence)

• Seasonal trends (Festivals, Elections, etc.)

Top 5 Most Crime-Prone Cities in India

top_cities = df["City"].value_counts().head(5)
print(top_cities)

Crime Trends Over Time (2000-2023)

import matplotlib.pyplot as plt
df.groupby("Year")["Crime Type"].count().plot(kind="line", figsize=(10,5), marker="o", linestyle="-")
plt.title("Crime Trends in India (2000-2023)")
plt.xlabel("Year")
plt.ylabel("Number of Crimes")
plt.show()

Step 4: Creating an Interactive Crime Heat Map for India

Using Folium to create a heatmap overlaying major Indian cities.

import folium
from folium.plugins import HeatMap

# Create base map centered at India

india_map = folium.Map(location=[20.5937, 78.9629], zoom_start=5)

# Convert data to list format

heat_data = df[["Latitude", "Longitude"]].values.tolist()

# Add HeatMap layer

HeatMap(heat_data, radius=10).add_to(india_map)

# Save and display map

india_map.save("india_crime_heatmap.html")

india_map

This heatmap will highlight crime-dense areas in Indian states and cities.

TOOLS AND TECHNOLOGIES USED:-

BENCHMARKING RESULTS:-

Using a dataset of 100,000 crime reports, we evaluated:

CONCLUSION:-

Crime data analysis using AI and visualization tools helps law enforcement and city planners make data-driven decisions. Our interactive heatmap can:

• Identify high-crime zones dynamically.

• Show temporal crime trends for better resource allocation.

• Help in crime prevention strategies using historical patterns.

📌 FAQ: Crime Data Analysis & Heatmap with FireDucks:-

1. Where does the crime dataset come from?

The dataset is sourced from: -

• NCRB (National Crime Records Bureau) → [https://ncrb.gov.in/](https://ncrb.gov.in/)

• Open Government Data (OGD) India → [https://data.gov.in/](https://data.gov.in/)

2. How does FireDucks improve data processing?

FireDucks is optimized for handling large-scale datasets. It offers:

• Faster operations than Pandas

• Efficient memory management

• Optimized geospatial data handling

3. Can this project scale to large datasets?

Yes! The project can handle large datasets by:

• Using FireDucks (better memory efficiency)

• Processing data in batches

• Using Folium & Plotly for interactive visualizations

For real-world applications, integrating BigQuery, Spark, or a database like PostgreSQL can help further.

4. How can law enforcement use this heatmap?

• Identify crime hotspots

• Allocate police resources efficiently

• Track seasonal crime trends

• Use AI to predict future crimes

5. Can I contribute to this project?

Yes! To contribute: -

• Fork the GitHub repository- https://github.com/angela479/crime-data-analysis

• Improve crime datasets, visualizations, and performance

• Submit a pull request

By integrating machine learning, future work could predict crime occurrences based on past data trends.

Google Colab Link: https://www.google.com/url?q=https://colab.research.google.com/drive/1XjcRXdHWR3YzyNNUFXfzLyKX8uyB3gH-?usp%3Dsharing&sa=D&source=docs&ust=1746627704755421&usg=AOvVaw0Nb_0L5X5Y4vo9m-P0HptK

GitHub Repository Link: https://github.com/angela479/crime-data-analysis