Tricky Traffic Tracker

Inspiration

I was inspired to do this project because I had recently gotten my driving license and was very afraid of car accidents. My mom, to this day avoids a certain intersection because I had a close relative get injured in a car accident. Had they planned the trip earlier, and decided where to go and where to avoid, perhaps things could have gone another way.

What it does

Our product tracks traffic live with API calls every thirty minutes and uses a RandomForestRegressor which recursively trains itself on its previous prediction and the last API call, making it a tracker for 20 intersections and roads in the Northern Virginia area. Concurrently, the model inputs data into a risk calculator which will provide a risk level of a certain intersection and also reroutes the user to go to a safer intersection that's less crowded at a certain time.

How we built it

I used a Django framework to organize all files into apps and sections that were specialized for the project into Python. The models would be Python, and Django is reliant on Python. We used the framework for a modern UI, functions like making an interactive map that you can click on and analyze, a login and register page, and a homepage. As for the RandomForestRegressor model, we used a dataset of 4000+ rows to train about 3200 rows, and test the other 800 with a 80/20 train/test split. Using Django made it really simple to link my dataset, API calls, and ML predictions into two different graphs that get compared against each other to train the model over and over again every thirty minutes.

Challenges we ran into

The biggest challenge we ran into was tiny errors in hoisting our final product onto Replit

Built With

• css3
• django
• html5
• javascript
• python
• randomforest
• sqlite