Car Accidents Insights

Based on Kaggle's "US Accidents (2016 - 2023)" Dataset

---

Dataset Overview

The dataset, titled "US_Accidents_March23.csv", contains information about car accidents across the United States up to March 2023. The data includes:

• Location details: City, state, and coordinates.
• Accident details: Severity, start and end times.
• Environmental features: Weather, road conditions, and more.

---

Objective

The primary goals of this project are:

1. Clean and preprocess the dataset for analysis.
2. Sample a subset of 10,000 rows for quick exploration.
3. Ensure datetime consistency for time-related columns (Start_Time and End_Time).
4. Perform initial analysis to highlight accident trends.

---

Steps Undertaken

1️⃣ Data Loading

The dataset was loaded into a pandas DataFrame:

import pandas as pd
file_path = r"US_Accidents_March23.csv"
df = pd.read_csv(file_path)

---

2️⃣ Cleaning Missing Values

To ensure clean and reliable data:

• Rows with missing (NA) values were dropped:

df_no_na = df.dropna()

This step ensures that only complete rows are used for analysis.

---

3️⃣ Sampling 10,000 Rows

From the cleaned dataset, a random sample of 10,000 rows was selected:

cleaned_accidents = df_no_na.sample(n=10000, random_state=42)

• random_state=42 ensures reproducibility of the sample.

---

4️⃣ Datetime Conversion

Columns Start_Time and End_Time were converted to datetime format:

cleaned_accidents['Start_Time'] = pd.to_datetime(cleaned_accidents['Start_Time'])
cleaned_accidents['End_Time'] = pd.to_datetime(cleaned_accidents['End_Time'])

This step ensures accurate handling of date and time values for trend analysis.

---

5️⃣ Exporting the Cleaned Sample

The cleaned dataset was saved for further use:

output_path = r"C:\Users\vange\Desktop\github project\cleaned_accidents.csv"
cleaned_accidents.to_csv(output_path, index=False)

• The cleaned file is saved at: C:\Users\vange\Desktop\github project\cleaned_accidents.csv

---

Team Members

This project was a collaborative effort by:

• Georgios Birmpakos
• Vasileios Katsikas
• Evangelos Diaskoufis

---

Questionnaire Highlights

📝 Key Areas of Focus

A. Accidents Per State

• Objective: Identify the states with the highest and lowest accident counts.
• Insights: Highlight accident trends across states.

B. Most Prone State for Accidents

• Determine the state with the highest accident frequency.
• Analyze if there's a significant gap compared to other states.

C. Accidents Per Month

• Goal: Examine seasonal trends in accidents.
• Identify peak months where accidents are most common.

D. Accidents Near Specific Locations

• Analyze accidents near:
  • Junctions
  • Stop signs
  • Traffic signals
• Calculate the percentage of total accidents occurring in these areas.

---

Next Steps

🔍 Exploratory Data Analysis (EDA)

1. Accident Severity Analysis:

   • Explore severity distribution across cities and states.
   • Identify the most dangerous areas based on accident severity.

2. Temporal and Spatial Trends:

   • Locate geographical hotspots for accidents.
   • Analyze time-based trends:
     • Peak accident hours
     • Seasonal patterns (e.g., months or holidays).

---

Summary

This documentation outlines the key steps in cleaning and preparing the "US Accidents" dataset for analysis. The cleaned dataset, containing 10,000 sampled rows, is now ready for:

• Deeper analysis of accident severity and trends.
• Visualization of hotspots, time-based patterns, and critical accident locations.

---

Stay tuned for further insights and visualizations! 🚗💥

---