• Boateng Gloria Maame Serwaa
• Dente-Quarshie Obeng Kwame

IMPACT: Led a team of three in pioneering the first-ever ML study of KNUST's student hostel market, utilizing regression and neural network models. Achieved over 77% accuracy, showcasing ML's impact on enhancing pricing transparency for students and stakeholders.

Machine learning, which dates back to the 1950s, aims to make accurate predictions with unseen (similar) data based on patterns discovered in existing data. In the real estate market, machine learning algorithms have been deployed in either determining house price indexes or estimating house sale prices. The latter is being studied the most by considering factors such as location, population, proximity to a nearby station, zip code, and many more. However, research on hostel price prediction is rare, if not nonexistent. To help bridge the gap, we will explore the impact of hostel features using three machine learning algorithms: multiple linear regression, ridge regression, and neural network, in predicting hostel prices. Empirical results support the potential of machine learning algorithms on the hostel market, with all $R^2$ greater than 0.75.

For this study, data were obtained through interviews with student residents in the hostels. The dataset is a cleaned dataset from 500 responses from 70 distinct hostels.
🚩 The accuracy values may be decieving due to the small number of data used in this study. On that note, we believe this is basically a baseline for future research.

• Charles Sebil PhD (Project Supervisor)
• Deborah Dormah Kanubala PhD
• Emmanuel Kofi Gavu PhD