Scribd
Upload
5 views11 pages

Crime Report and Analysis System

The document outlines a project focused on using machine learning to predict crime trends based on historical data, aiming to enhance law enforcement actions. It details the dataset, preprocessing steps, model architecture, and evaluation metrics, emphasizing the integration of live crime reports for real-time analysis. Future improvements include implementing advanced models, utilizing granular data, and enhancing visualizations for better decision-making.

Uploaded by

anannyaaa98
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
5 views11 pages

Crime Report and Analysis System

The document outlines a project focused on using machine learning to predict crime trends based on historical data, aiming to enhance law enforcement actions. It details the dataset, preprocessing steps, model architecture, and evaluation metrics, emphasizing the integration of live crime reports for real-time analysis. Future improvements include implementing advanced models, utilizing granular data, and enhancing visualizations for better decision-making.

Uploaded by

anannyaaa98
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 11

Crime Report and

Analysis System
Machine Learning for Crime Trend Prediction

Presented by: Anannya Anesweta Rout

Regd. No.: 2101413023

8th Semester Project


Role of Machine
Learning in the
Project
Objective
• Forecast future crime using historical data
• Enable data-driven law enforcement actions

Key Benefits
• Early high-risk area identification
• Visual insights for officials and public
• Live crime report integration
Dataset Overview
Primary Source Data Fields Supplementary Data

The dataset used is Location: Geographic information In addition to the static CSV dataset,
crime_dataset_india (1).csv, which such as city, district, and state where this system integrates live crime
contains comprehensive crime records the crime occurred. reports captured in real-time via a
compiled from various official sources Crime Type: Categorization of the MySQL database.
across India. crime, including theft, assault,
It includes approximately 10,000 This enables updating the analysis and
burglary, cybercrime, and more.
individual crime records spanning from predictions based on the most current
the year 2010 through 2023, Date of Occurrence: Timestamp of incidents, facilitating timely law
capturing trends over a 13-year period. when the crime was reported or enforcement responses and
happened. community awareness.
This extensive temporal coverage
Additional Fields: Some records
allows for longitudinal analysis and
include victim demographics, case
trend detection in criminal activity
status, and reporting agency.
across different regions.
Data Preprocessing Steps
Date Conversion Encoding
Transformed to Label encoded Crime
year/month format for Type and Location fields
uniformity

Data Grouping & Cleaning


Grouped by year, location, and crime type; removed
duplicates and nulls
Feature Engineering
Time Series Creation Feature Set
Data was grouped by year, The final set of features
crime type, and location to includes Year, Crime Type,
create meaningful time series Location, and Crime Count.
features. This grouping allows These features were carefully
for identifying patterns and selected to capture temporal,
trends over time within specific categorical, and spatial aspects
areas and crime categories, of the crime data to enhance
which is crucial for predictive model accuracy and
modeling. interpretability.

Normalization
Numerical values such as Crime Count were normalized to a standard
scale to ensure compatibility with machine learning algorithms. This
step helps improve the model's convergence speed and performance by
preventing features with larger scales from dominating the training
process.
Model Architecture

Algorithm Reasons Inputs & Outputs


Linear Regression models crime Simple, interpretable, efficient, Features: Year, Location, Crime Type;
trends over time for forecasting robust against overfitting. Output: Predicted crime count.
future occurrences.
Model Training & Evaluation
Training Setup
The dataset was split 80% for training and 20% for testing, enabling effective
learning and unbiased evaluation. Keeping test data separate reduces
overfitting risk and ensures realistic performance assessment.

Validation
5-fold cross-validation enhanced robustness by training on four subsets and
validating on one, reducing bias and ensuring consistent results across data
samples.

Metrics
Performance was measured using Mean Squared Error (MSE) and an R² score
of ~0.87, reflecting strong prediction accuracy. Monitoring these metrics
helped detect overfitting or underfitting for timely adjustments.

Tools
Python libraries like Scikit-learn, Pandas, and NumPy were used for efficient
and reliable model training and data processing.
System Integration

Model Deployment Backend


Saved the model as a linear_model.pkl file Flask serves predictions through a REST API
Frontend Visualization
Tools Used Interactive Filters Purpose

Chart.js and Plotly • Location: Allows The visualization


libraries are utilized to users to filter crime interface displays
create dynamic and data by specific projected rises or
visually engaging charts geographic areas for declines in crime trends
and graphs. These tools more precise based on predictive
enable real-time • insights.
Crime Type: Enables modeling. This helps
updates and responsive selection of different stakeholders and law
designs suitable for categories of crime enforcement agencies
various devices, to analyze trends to identify high-risk
enhancing user individually or areas and timeframes,
experience. comparatively. facilitating proactive
measures and strategic
• Time Period: Users
planning.
can adjust the time
range to examine
crime trends over
days, months, or
years.
Future Improvements
Advanced Models
Implement LSTM and Prophet for improved accuracy.

These models can capture temporal dependencies and seasonal trends more effectively
than current methods, leading to better crime trend predictions.

Granular Data
Use district-level crime location data.

This more detailed data will allow the system to detect localized patterns and hotspots,
enhancing predictive precision for specific neighborhoods.

External Factors
Incorporate population and event data.

By considering demographic shifts and planned events, the model can account for
external influences on crime rates, improving forecasting robustness.

Visualization Enhancements
Add geographic heatmaps for spatial analysis.

These interactive heatmaps will help users visually identify high-risk areas and
temporal changes in crime, supporting better decision-making by stakeholders.
Conclusion Highlights

Effective Crime Trend Prediction Seamless Real-Time Integration Interactive Visual Forecasts
Robust machine learning models tailored for Continuous updates from live backend data sources. User-friendly charts offering actionable insights.
forecast accuracy.

Foundation for Smart Policing Path Forward


Combining data science and real-time analytics for Planned upgrades for greater accuracy and usability.
safety.

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy