Alberta Collision Report

heat map of car collisions in Alberta

Inspiration

Our goal is to raise awareness on road safety, specifically in Alberta. Currently, there is no publicly accessible compiled information on car collision locations. As well, nobody else analyzed the trends in car collision using machine learning in Canada.

What it does

It compiles information through web scraping, analyzes government data on car collisions through machine learning, and displays above data in a website.

How we built it

To compile information through web scraping, we scraped news report data from a website whose approach is one of the most systematic and thorough in recording locations of collisions (https://www.navbug.com/alberta_traffic.htm). To analyze government data on car collisions, we used two machine learning training models -- ARIMA and Prophet -- to analyze trends and to predict future occurrences.

Challenges we ran into

We were new to all the techniques we used -- web scraping, machine learning, and web designing. However, each of us were motivated and trained really hard within a 24-hour time frame to put together a meaningful, complete project. For this reason, unfamiliarity with what we can do with python was the main challenge we ran into.

Accomplishments that we're proud of

Originality

Currently, there is no publicly accessible compiled information on car collision locations.

Complexity

As well, nobody else analyzed the trends in car collision using machine learning in Canada. We are proud that we took on this project in an original manner after quickly learning advanced python techniques, such as web scraping and machine learning.

Execution and Polishness

We completed the working project within 24 hours.

Utility

Our project raises awareness on road safety and can be extended to encourage the government authorities to implement our system through showcasing its importance.

What we learned

We learned how to efficiently work collaboratively through the GitHub.
As well, we learned that we should not overestimate the amount of time we have left until the deadline. Instead, we need to always work diligently even if there appears to be a lot of time.

What's next for Alberta Collision Report

Since scraped data had a little bit of irrelevant and redundant information, we want to incorporate machine learning into the scraping techniques to prevent such information from being scraped.