The U.S. census provides us with important information regarding our nation’s demographics. Yet, it fails to account for a frequently overlooked group of people that constitutes 18% of our population: the homeless population. Given the mobile nature of homeless populations, it is difficult to keep track of their movement despite many governmental attempts to do so, which makes providing support a difficult task. The goal of our project, Invisible Populations, is to pinpoint areas of high homelessness through various predictors and determine optimal locations for the construction of shelters or supportive establishments. Ultimately, we envision our project to be used by non-profit organizations in determining which areas are likely to require homeless relief and efficiently allocate their resources.
Numerous studies have identified several risk categories such as demographic characteristics, population density, cost of living, physical/mental health status, substance use, involvement with the criminal justice system, and housing conditions as indicators of high rates of homelessness. Through extensive research, our team selected 5 main factors that correlated with homelessness. After scouring the internet, we found large datasets that displayed the prevalence of each risk factor in the U.S.. We weighed each risk factor for the likelihood of contributing to homelessness and through statistical analysis, and we combined the datasets to form a model that more accurately represented the geographical locations of homeless populations. Our model is a progressive step towards a comprehensive locator for areas of need.
Although our primary goal is to get unsheltered populations the support they need through nonprofits, we hope that our project will also be used by the public to become more educated on the topic of homelessness. Our project achieves this with an informative and interactive UX design. Our easy to navigate website features a slide carousel with insightful facts and provides supporting readings in hopes of spreading awareness. Additionally, with the help of Velo by Wix, we created a form that allows users to search the database of information that we compiled per county, to allow the public to access valuable information about homeless populations in a given area. Through both our Velo-powered location search tool and our geographical model, Project Invisible Populations presents an educated prediction on the counties that are at high risk for increased levels of under supported homeless populations.
Our team faced many challenges throughout the development of our visualization tool. We picked a complex area to model, so we had to conduct research to support our methods and conclusions. Finding which demographic and population information was important was crucial, and finding the supporting data sets took a significant amount of time.
Once we did the initial scoping, research, and data collection, our three main work areas were creating a way to visualize the data, analyzing the data meaningfully, and making the website. In order to visualize the data per county, we had to convert all of the data to county-specific, which was difficult given that much of it was in city form, which we had to convert to county-level instead. To do this, we located the geographic location of the city, and we used a mapping API to find the longitudinal value of each city to map to counties. This created issues like having to input much of the county data manually due to the geography not syncing, and some states not having direct city data, which forced us to go through and map many locations by hand. Additionally, some sources completely omitted county data for a couple states which impelled us to write a script that scrapes governmental lists to create a dictionary from towns to cities. Also, creating our index of overall risk per county was contingent on fully understanding the factors creating higher homeless incidences.
In terms of the website, implementing the per-county search was a steep learning curve, and formatting appropriately given our large amount of research and supporting information was time-consuming. Given the sheer amount of factors that we compiled about this important issue, creating digestible and useful content was imperative and challenging. We learned a lot, both about our topic and a broad variety of skills, including Wix design and a lot of data analysis techniques.
Given the time we had, we were only able to take five factors into account, but in the future, we plan on expanding our project and making our geographic model more accurate through the implementation of other decisive factors of homelessness. We would extend our data to include factors like race, and we would see if we could directly work with the government to get internal information to synchronize with our compiled information.
Log in or sign up for Devpost to join the conversation.