Traffic accident frequency based on weather conditions and age group.

Datasets: Traffic accident data, weather data, individual data.

Data 1 : NYC ( Motor Vehicle Collisions - Crashes) (https://data.cityofnewyork.us/Public-Safety/Motor-Vehicle-Collisions-Crashes/h9gi-nx95/data )

Data 2: NYC open-meteo.com (extracted as CSV)

Data 3: NYC (Motor Vehicle Collisions - Person) https://data.cityofnewyork.us/Public-Safety/Motor-Vehicle-Collisions-Person/f55k-p6yu/about_data

Github files: excel sheets (all three unclean and clean datasets), model .ipynb and .py files.

Questions seeking to answer :

1. Are there seasonal patterns in accident occurrences, and do they correlate with specific weather conditions? Is there any relationship between weather conditions and collisions?
2. Predict the likelihood of an accident based on time of day, season, and weather conditions.
3. Are certain age groups more likely to be involved in accidents under specific conditions? Identify high-risk age groups. Can we design targeted safety campaigns for high-risk age groups? We can classify the age group into seniors, teenagers, or maybe by generations (Genz, millennials and all) Any relationship between the type of vehicle and the frequency of the accident?

Dataset Variables

This is a list of variables included in the original export. The final version of the cleaned dataset is available in this repository.

Data 1 (Collision Data)	Data 2 (Precipitation)	Data 3 (Demographic)
Crash date	Highest/Lowest Maximum Temperature	UNIQUE_ID
Crash Time	Highest/Lowest Minimum Temperature	COLLISION_ID
Borough	Highest Precipitation	CRASH_DATE
Zip Code	Highest Snowfall	CRASH_TIME
Latitude	Highest Snow Depth	PERSON_TYPE
Longitude	Date	PERSON_ID
Location		PERSON_INJURY
On street name		VEHICLE_ID
Number of persons injured		EMOTIONAL_STATUS
Number of persons killed		BODILY_INJURY
Number of pedestrians injured		POSITION_IN_VEHICLE
Number of pedestrians killed		SAFETY_EQUIPMENT
Number of cyclist injured		PED_LOCATION
Number of motorist injured		EJECTION
Number of motorist killed		PERSON_AGE
Contributing factor 1

Data Extraction

• Extracted datasets are CSV files from various websites.

• Streamline the first 5 years to reduce the dataloading (2019 - 2023). 12am 31st Dec 2018 - 6:18pm 11 Nov 2024. This was the time at which the data was extracted.

Data Cleaning

• Date and time column formatting.

• Delete columns and fill missing data with n/a. (see the notebook for cleaning steps)

• Limit the datasets to only 5 years.