This project is a comprehensive Exploratory Data Analysis (EDA) on the Los Angeles Crime Dataset (2020–Present) using Python, Pandas, NumPy, Matplotlib, Seaborn, and Scikit-learn.
The goal is to uncover trends, patterns, and insights about crime incidents using real-world data.
- Source: data.lacity.org
- Name:
Crime_Data_from_2020_to_Present.csv - Records: 1,000,000+ rows
- Features: Date, Time, Location, Crime Type, Victim Descent, Weapon Used, etc.
| Category | Tools / Libraries |
|---|---|
| Language | Python 3.x |
| Data Manipulation | Pandas, NumPy |
| Data Visualization | Matplotlib, Seaborn |
| Statistical Analysis | Scipy (Shapiro-Wilk, Chi-squared) |
| Machine Learning | Scikit-learn (Random Forest Classifier) |
| IDE | Jupyter Notebook / VS Code |
1_Data_Loading_Cleaning.ipynb→ Load CSV, clean missing/nulls, handle datatypes2_Visualization_EDA.ipynb→ Univariate, Bivariate plots (Time, Location, Crime Type)3_Statistical_Analysis.ipynb→ Correlation, Outliers, Hypothesis Testing4_Machine_Learning_Model.ipynb→ Simple Random Forest Classifier to predict crime typeREADME.md→ Project Overview
- 🔥 Peak Crime Hours: Most crimes happen between 6 PM to 11 PM.
- 🗺️ Hotspots: Downtown LA and South LA see maximum criminal activity.
- 🕒 Monthly Trends: Crime frequency dips during pandemic lockdown periods.
- ⚔️ Weapons: Highest number of crimes committed without weapons or with handguns.
- 📊 Classification Accuracy (RandomForest): ~72%
- Real-world data cleaning & type conversions
- Data visualization principles and customization
- Statistical testing using Python (Shapiro, Chi-Square)
- Simple ML model training and evaluation
- Importance of feature engineering in time/location-based datasets
- Add clustering for unsupervised crime pattern detection
- Deploy model using Flask or Streamlit
- Create an interactive dashboard with Plotly or PowerBI
git clone https://github.com/kunalusername/Crime-EDA.git
cd Crime-EDA
jupyter notebook