Case Studies

The City of Las Vegas uses predictive modeling to pinpoint vacant properties and proactively monitor for squatters and fire risk

Background

The City of Las Vegas, Nevada, underwent rapid growth throughout the latter half of the 20th century, and the population nearly doubled in the decade from 1985 to 1995. Expansion continued through the early 2000’s, but the 2008 financial crisis hit Las Vegas particularly hard. Home prices fell 62% from their peak in 2006, and single-family building permits declined over 90% (source). Although the tourism sector has partially rebounded, and home prices saw double-digit growth from 2017 to 2018, the City continues to experience above-average rate of vacant homes. In 2017, data from the US Postal Service indicated over 14,000 homes, or 2.17% of the overall housing stock, were vacant. And while the national average is 1.58% and declining, Las Vegas’ vacancy rate is increasing. As a result, Las Vegas Police receive over 5,000 calls annually about squatters, and vacant property fires are a regular occurrence (source).

The Challenge

For Vicki Ozuna, Code Enforcement Supervisor, building a reliable inventory of vacant properties was a critical first step to monitoring and safeguarding these high-risk buildings. “We can’t just wait for a neighbor to complain—or worse, for a fire to break out or someone to get hurt. But we’ve got over 200,000 properties in Las Vegas, and we can’t put eyes on every single one.”

Getting data on what properties are and are not vacant proved to be a difficult task. A registry of vacant and foreclosed properties relies on voluntary compliance from owners. Some utilities do not require verification that the person requesting service is an authorized occupant of the home. And Ozuna’s team of code enforcement officers used different methods to flag potential vacants within their case management software. As a result, the Code Enforcement Department kept an official list that tallied only 150 known vacant properties—a number that Ozuna knew to be a fraction of the real total in the City.

By analyzing the common characteristics of known & confirmed vacant buildings, then identifying other properties with similar characteristics, Tolemi’s data scientists were able to point Ozuna’s team to her most likely candidates for a proactive inspection. Factors such as the words “vacant” or “abandoned” in the notes field of police, fire, or code enforcement reports; tenure of open violations; ratio of delinquent taxes to assessed value; length of time since last water utility payment; and census block group vacancy rate all contribute to a high vacancy risk score. To operationalize this data, Ozuna simply opens the BuildingBlocks application, filters for properties with a score of 90 or above, and filters out any known vacants. In a single click, she is able to segment these properties by code enforcement area to assign them to her officers, who then inspect each one and record its vacancy status, which is fed back into BuildingBlocks.

In her initial trial run using the vacancy risk score, Ozuna identified the 50 properties that scored highes for proactive inspections. The result: her officers found that 46 of them were vacant but previously unreported. “We were blown away,” she reported. “So obviously we stepped up our efforts.” Within a few months of launching BuildingBlocks, Ozuna’s list of known vacant properties grew from 150 to over 5,000. “Now we have a grasp of the scale of the problem, as well as the ability to pinpoint specific properties that need our attention. There are implications for nearly every department and agency in the City. Fire, Police, Housing, even Education can use this information to better serve our citizens.”

“It’s not a question of how long it would have taken us to get here without BuildingBlocks. There is simply no way we could have uncovered all of these high-risk properties without this tool.”