Really Epic Team 05 — Project 1 Proposal

Project Title: Econometric Property Investment Calculator (EPIC)

Team Members:

• Troy Bailey
• Seth Bitney
• Katarina Pin
• Valerie Wilmot
• Yuta Yamaguchi

Project Description/Outline:

We will analyze property values in the greater Austin area as they relate to potential influential factors using heat mapping as the primary visualization form.

Research Questions to Answer:

How are property values influenced by different factors such as commute, crime rates, school districts, population density, availability of healthcare resources, variety of restaurant choices, dog ownership, etc.?

Data Sets to be Used:

• Zillow
• OpenStreetMap
• City of Austin Census
• US Census Bureau
• School Digger
• Texas School Guide

Data Acquisition and Cleanup

Started with 40,000 randomly generated coordinates with within a 20 mile radius of the Texas State Capital, then converted to physical addresses classified as residential homes (20,177). From there, we dropped duplicates and ran addresses through the Zillow API using the PyZillow module to return values for zEstimate and square footage. Properties without values were dropped, as well as properties with unreasonable values (e.g. $9M with 1 sqft) resulting in 8,530 unique properties (see code in <EPICdata.ipynb> for details on parameters set). We then collected commute times for each address based on census data to create commute scores, crime data across the area evaluated to assess safety scores, and school ratings based on the nearest public school to each address.

Findings

Heat maps were created for each of the variables evaluated and are shown below.

Valuation

Property values were assesed based on value per square foot (zEstimate / square footage). Results displayed as darker red indicating higher value per square foot.

Commute Score

Commute score results displayed where darker blue indicates shorter commute times.

Safety Score

Safety score results displayed where darker blue indicates safer areas.

School Rating

School rating results displayed where darker blue indicates higher school rating.

Using the preceding heat maps, we overlayed the maps to show interactions between valuation and each of the variables evaluated.

Value & Commute Score

High property values and low commute times align with locations of big companies.

Value & Safety Score

With the exception of downtown, high property values correlate positively with high safety scores.

Value & School Rating

Property values and school ratings do not have a strong relationship based on the data used. There many be confounding variable not assessed, such as private schools or areas where the neighborhood is converting from low income to high income and the schools have not yet caught up.

Further Analysis

In future iterations of this project, we will asses total zEstimates, lot sizes, and value per lot size in relation to these and other variables including walk score, access to healthcare, access to services such as restaurants and stores, as well as other measures. We will attempt to account for all the variability in housing prices with the variable assessed to ultimately predict property values based on input values given.

Rough Breakdown of Tasks:

• Define questions to ask
• Task management system setup on Trello - Kat
• Assign github owner - Valerie
• Create documentation and setup files - Valerie
• Research different datasets - Troy, Seth, Kat, Yuta, Valerie
• Scheduling time for meetings
• Exploration of data - Troy, Seth, Kat, Yuta, Valerie
• Cleanup of data - Troy, Seth, Kat, Yuta, Valerie
• Analysis of data - Troy, Seth, Kat, Yuta, Valerie
• Create presentation with Google Slides - Kat & Valerie
• Double check presentation requirements - Valerie
• Double check presentation guidelines - Valerie