Original Project Proposal

Project Details

Data Sources:

• Accident Data: https://www.kaggle.com/daveianhickey/2000-16-traffic-flow-england-scotland-wales/
• Weather: OpenWeatherMap API
• Original Data: https://discover.ukdataservice.ac.uk/catalogue/?sn=7752&type=Data%20catalogue#documentation

Libraries:

• Pandas
• Seaborn
• Matplotlib
• Plotly
• Numpy
• scikit-learn
• OS

Data Import & Cleaning: Charles!

Statistics

For all stats questions, we will say results are statistically significant at:

• 95% Confidence Interval / P-Value < 0.5

Questions:

1. Correlation between weather and severity/casualties? (Pratham)
2. Correlation between time of day and accidents? (Dolly)
3. More/less casualties based on road type (taking into account urban vs rural)? (Dolly)
4. Correlation between accident COUNT and day of the week? (Heidi)

Features:

• Basic visualizations (Heidi)
  • map viz
• Create a predictive model using Naive Bayes relating weather conditions and the likelihood of an accident. (Pratham)
  • Predict likelihood of getting into an accident based on weather conditions

Roadmap (Due Sun. March 18th)

1. EOD Sun March 11 (Ambitious)/EOD Mon March 12 (Final): Have cleaned dataset uploaded to Github
2. By Start of Class Thurs March 15: Have at least one visualization done per question
3. By Start of Class Sat March 17: Have everything done
4. By 6PM Sun March 18: Review and turn it in