Volunteer data scientists have converted 5 years of data about sheltered homeless people in Downtown San Diego from hand-marked paper maps to a detailed geographic dataset, using a combination of machine vision tools, Python programs and manual work. This talk will review how we created the dataset and how we are working with government and universities to use the dataset to inform social policy.
For the last 7 years, the Downtown San Diego Partnership has been conducting monthly counts of homeless in the Downtown neighborhood. The data is recorded on paper maps, which are compiled into a spreadsheet. This is a fantastic dataset which would be even more useful if the hand-recorded paper maps were digitized. This project involves collecting and digitizing 5 years of the monthly maps, to produce a geographic dataset, whcih we will analyze for time trends and for the association between homeess movements, geography and the built environment.
This talk is suitable for people with an interest in using data to solve social issues any level of skill.
This talk will:
- Describe the scope of the homeless problem in San Diego and present the opportunities for data science to inform homeless policy.
- Detail the process we used to convert the hand-marked maps, both the manual process and use of machine vision
- Show the analysis of the final dataset, with specific emphasis on geographic analysis using Geopandas and Jupyter.
- Demonstrate the data management process, using Metatab to package data and publish it to Wordpress
- Present how we are training volunteer analysts in the PyData tools to answer data questions posed by researchers.
Attendees will learn important aspects of an important social problem and how to use data to solve these problems using a range of Python tools.