FareCheck (Team 7)

Inspiration

We aimed to create a more reliable, accessible, and data-driven way for travelers to make travel plans, particularly for budget-conscious individuals such as students.

What it does

We developed a model that predicts prices and provided an interactive web interface that allows individuals to select two locations and receive a prediction along with a user-friendly transparency behind price differences, such as whether the area is a premium hub or not.

How we built it

We built an XGBoost regression model that predicts U.S. airfare using distance, demand, competition, and hub characteristics to capture real economic pricing dynamics. We deploy the trained model using Flask, which serves an HTML interface where users can input route characteristics and receive real-time price predictions from the model. Finally, we use SHAP to explain each prediction, showing which features drive the fare up or down so the system is both predictive and interpretable.

Challenges we ran into

We encountered several challenges during the development of the website, including difficulties in excluding confounding variables for data analysis, integrating separate models into the final website, and finding reliable external datasets to support our conclusion. These challenges were addressed through collaborative and proactive problem-solving, leading to improved reliability and functionality.

Accomplishments that we're proud of

We're proud of the successful completion of the website with the addition of SHAP to quantify and visualize the impact of feature contributions, overcoming a multitude of challenges along the way. Developing the prediction model was a huge accomplishment, as this was a new challenge for us and pushed us to learn more about the properties and usages of machine learning methods such as XGBoost and Random Forest.

What we learned

In developing FareCheck, we learned the importance of collaboration and persistence in problem-solving when working with complex datasets. We gained valuable insights into the sanitation of data and the diverse conclusions that can be found, as well as the challenges of creating a reliable and user-friendly website. We also learned to adapt to unforeseen difficulties, like technical setbacks and design flaws, and use them as opportunities for growth.

What's next for FareCheck

We want to expand beyond the provided dataset and highlight the underlying socioeconomic factors in the airfare market mechanics. By analyzing the relationship between price and variables such as GDP growth and education, we can uncover affordability and inequality within the route-level prices, providing policymakers with the ability to translate into actionable change.