One of Los Angeles' most prominent and publicized concerns is the homeless population. According to Los Angeles Homeless Services Authority, the county has over 58,000 homeless. There have been several initiatives in attempt to decrease the number, including the Affordable Housing Trust Fund started in 2003. However, despite the Los Angeles Housing + Community Investment Department (HCIDLA) building almost 400 projects up until now, the numbers only seem to be rising. If Affordable Housing isn't helping, more attention should be allotted to alternate solutions or changing the projects. By mapping homeless communities reported through the 3-1-1 line of Los Angeles County in 2015 to 2018, a reliably accurate trend can be determined of where they are located. By overlaying these maps with the maps of locations of affordable housing units built that year, it can be determined if these projects have an effect on the homeless population around them. I used Python's panda to sort through the data from LA's open data website of homeless encampment calls in 2017. Then using Google V3's API in the geopy library, it took the addresses of the calls and turned them into longitude and latitude values. I complied the csv file into a geojson file for maps to read. Finally, using the javascript library Folium's Leaflet, I created an interactive map on website with all the coordinates plotted. Due to the huge amount of data, unfortunately, the map was super slow and couldn't properly display all the data points. There is an extension of Leaflet that allows the clustering of data, but I couldn't figure out how to add it properly. I think I've learned a lot about Big Data and the implications it can have simply by the research I had to conduct to come up with this idea. Additionally, I struggled a lot because I had not ever used any of these libraries and many of the tutorials online were not for such large amounts of data. It's pretty amazing however though I could process almost 3000 data points with a few lines of code (that took hours to tweak). In the future, it would be great if I could implement the clustering to make the UI more friendly and the overall program more functional. If this works, I could add overlay all the maps to see the trends at the same time. Adding a scroll bar to see the populations change over time would be pretty interesting as well! (best domain)

Built With

Share this project: