AI-Powered Infrastructure Demand Forecasting

An example of the output (based on user input cities)
Project Overview
Business & Social Impact

Inspiration

Infrastructure investments like airports often rely on outdated data, while tourism demand shifts much faster. We were inspired to build an AI tool that forecasts where travel interest is growing in real time, aiming to help policymakers and investors make smarter and data-driven decisions before congestion or inequality arise.

Our project have business and social impacts. In terms of business impact, it supports investors to identify high potential airport locations, optimize return on investment (ROI), reduce risk of over or under investments, and foster collaboration between public and private sectors. From a social impact perspective, our project promotes tourism development, boosts local economies, and improves accessibility by identifying underserved regions, which helps governments reduce regional inequality and stimulate job creation.

What it does

Our model captures global travel trends from social media (Wikivoyage), extracts sentence level sentiment and mentions, and ranks United States cities (user input) based on rising tourism demand. The result shows a ranking where new airport infrastructure would generate the highest social and financial impact.

How we built it

We built an end-to-end pipeline that automatically collects and analyzes city-level (USA) travel sentiment from Wikivoyage. The system first crawls Wikivoyage pages using the MediaWiki API and BeautifulSoup, extracting and cleaning key travel-related sections such as “See,” “Eat,” and “Stay safe.” Emotionally expressive English sentences are prioritized and saved as structured text data. Then, using VADER sentiment analysis, the pipeline computes per-city sentiment metrics, such as positive ratio, average sentiment, and an overall composite score, to rank cities by traveler sentiment. Finally, an interactive interface allows users to input three cities and instantly visualize their sentiment-based rankings.

Challenges we ran into

We faced challenges when trying to crawl data from social medias, including Reddit, Twitter, and Wikivoyage because of API related issues. We also faced issues when trying to build a user interactive interface. However, we overcame them (but it was very hard for sure)!

Accomplishments that we're proud of

This is our first ever Hackathon, so we are very proud because we actually came up with an idea on the spot, collaborated, and built a project that works! We also had fun building this project and using what we learned in the MAIS 202 bootcamp.

What we learned

We learned how real time text data reveals tourism trends faster than traditional dataset. We also are very proud to learn how to build an end to end pipeline.

What's next for AI-Powered Infrastructure Demand Forecasting

For next, we want to expand the dataset to include Reddit and Twitter for real time travel sentimement. And we want to build an interactive visualization dashboard or an app for policymakers and investors. We may also integrate economic indicators to help support decision making.