Student project predicts traffic before it happens | St. Clair College
The Windsor-Detroit Traffic Forecaster is a free-to-use web application that uses data gathered from the City of Windsor’s own records to predict the amount of traffic in the Windsor-Detroit border crossing. It uses artificial intelligence to help gauge these amounts.

St. Clair College students have created a system to predict local traffic.

The Windsor-Detroit Traffic Forecaster is a free-to-use web application that uses data gathered from the City of Windsor’s own records to predict the amount of traffic in the Windsor-Detroit border crossing. It uses artificial intelligence to help gauge these amounts.

It was designed by a trio of Data Analytics students comprised of Luis Silva, Julia Kumala and Harish Veeramosu. Group member Luis Silva said the application will assist users in a variety of ways.

“We look to help residents and businesses to have smart transportation planning, when they have insights like forecasting traffic conditions, volume expected in future dates and intelligence advice for planning,” said Silva.

Silva also said the project was designed so users don’t need to have a background in statistics to understand the predictions.

The process of designing the app came down to a couple of stages. The first priority was to brainstorm solutions to transportation issues.

“We identified there was a need on this matter for business which look to improve international shipments and residents who commute in the border area. This was also highlighted by the Canadian and American governments having their priorities to improve traffic,” said Silva.

From there, Silva said the group moved on to gathering and analyzing data related to traffic problems. Part of this process was reliant on what Silva calls EDA or Exploratory Data Analysis. It was here they gathered data from the City of Windsor on border crossings.

The group then used two techniques, called feature engineering and feature importance. These techniques allowed the group to identify all the features their final product should offer to users.

What followed was a lot of experimenting with various models, or ways of working with the data they gathered previously.

“We looked to evaluate performance metrics, bias and conclude which model-setup predicts the best for this data,” said Silva.

This process involved more than 100 attempts at finding the right setup to predict Windsor and Detroit border traffic, according to Silva.

Finally, the group adapted their information to a web application.

It was presented at the ONE Ford Capstone event this past Tuesday, where it ranked third out of the 16 projects shown. It also won the team a $500 cash prize.

The accessibility of the application online plays a big role in how it can benefit the community.

“This web application is available to everyone who has access to the Internet and is looking to forecast traffic conditions in the Windsor-Detroit border, for personal or business interest,” Silva said.

- Tyler Clapp