Inspiration
Because of the COVID-19 pandemic, so many areas of our lives have drastically changed, notably our willingness and ability to travel. Since our team believed that previously existing traffic models may not be as accurate, we sought to create a projection of traffic more relevant to our current situation.
What it does
KovidTrafik takes a user input of a date and time. The machine learning algorithm behind KovidTrafik then predicts the traffic level at the inputting time and prints out whether traffic is high, moderate, or lower, as well as how the traffic level compares to the average traffic levels.
How I built it
The machine learning algorithm behind KovidTrafik was built with an ARIMA time series model through python. The website was made on IBM cloud and formatted through html and css.
Challenges I ran into
Because this was our first experience with machine learning and hackathons in general, some difficult concepts bewildered us. Visualizing our data and creating a usable website were both new and challenging to us.
Accomplishments that I'm proud of
We are proud of the accuracy of KovidTrafik. We believe the model predicts values within a range implementable in the real world. The website we created was also more complex than our previous creations.
What I learned
We learned the concepts of machine learning, specifically ARIMA time series. Specifically, lag order and degree of differencing were ideas that required us to spend time on to comprehend.
What's next for KovidTrafik
We aspire to compile even more data for KovidTrafik, making the model more accurate. By increasing its accuracy, we hope to make KovidTrafik usable and beneficial around the world.
Log in or sign up for Devpost to join the conversation.