In these sample papers from the 2024 M3 Challenge, students in the U.S. and the U.K. were tasked with using mathematical modeling to understand the intertwined housing and homelessness crises from a long-term perspective. Each paper has a cover sheet that indicates what judges liked about the paper and what could have been improved.
“Our findings lead us to one conclusion: to tackle homelessness in places like Seattle, we need to tackle affordability issues first and foremost. The large amount of available housing units that go unutilized prove simple lack of housing isn’t the main problem, and natural disasters have surprisingly little correlation with the homeless population. While addressing either of these issues could be beneficial, funding is best allocated to addressing the core of the problem: the thousands of functional but unused housing units and the financial barrier that stops those in need from accessing them. Housing vouchers have proved a successful solution in many places that use them; if we can focus on evening out the inequality between income and house prices in the long term, we can make it easier than ever for the homeless to find a place to call home.”
“First, we predict future vacant unit amounts across Seattle, Washington, and Albuquerque, New Mexico by using Holt-Winters, a time series forecasting model, specifically designed for capturing and predicting patterns in data that exhibit specific trends…Next, using GRU (Gated Recurrent Unit), a neural networking model involved in time series forecasting we extrapolated the pattern and applied it to the proximal future…Thereafter, we used the same model again, GRU, to predict the homeless population after the solutions were implemented…We believe that through these results, policy-makers and leaders in major industries will be able to gain insight on how to assist the growth of the economy and prevent major damage to the housing market.”
“We predicted the housing supply in cities in the UK in 2034, 2044 and 2074 using an amalgamation of multiple uni-variate logistic regression models…To further reinforce our study, we followed with a consideration of how homelessness will rise over the same spans of time, and in the same UK locations…To corroborate with your findings, we have conducted a smaller scale investigation, examining 3 different actionable schemes (Building new flats, increasing the minimum wage and increasing rent on social houses) and also the effect of migration on our homeless population, through a combination of various simple techniques – this should provide a good starting point for a more in-depth future study into these issues. However, all policies show promise and have varying cost-to-benefit ratios, with increased social rent demonstrating a potential halving in homelessness.”
