Some are skeptical, hopeful of APD’s ‘machine learning’ approach to staffing shortage


AUSTIN (KXAN) — Rebecca McKee’s North Central Austin upholstery shop was hit by thieves back in October.

“They stole everything from our officer, laptops, paperwork, blank checks,” she told KXAN.

McKee said she had a hard time getting help from police, so she was glad to hear computer algorithms have given Austin’s police chief a new tool to decide how many patrol officers it really needs.

“It’s a good first step,” McKee said. “I mean, it’s pretty obvious that we are lacking police staff right now.”

As KXAN reported, a report put commissioner and funded by the Greater Austin Crime Commission utilized a machine learning model to sift through five years worth of police call data. The model recommended an additional 108 patrol officers. The department is currently authorized for 774.

So, what is machine learning?

To understand the technology, we turned to the experts at the University of Texas at Austin’s Machine Learning Laboratory.

When asked the simplest way possible to explain what machine learning is, director Adam Klivans said, “It’s an automated system that makes intelligent predictions based on data that it’s seen in the past.”

“One of the easiest examples would be to think about your email program that you use,” he said. “Every time that you get an email, whatever system you’re using, it first decides, ‘Is this a spam email or not?’ There has to be some algorithm or some decision-making that goes on, based on other emails that [it] has seen in the past to figure it out.”

Laboratory co-director Alex Dimakis warned the outcomes are only as reliable as the data provided.

“Machine learning models are so complicated that trusting their predictions, you have to be very careful,” Dimakis said.

KXAN also reached out to Kathy Mitchell with the bipartisan criminal justice reform group Just Liberty. Mitchell also had a seat at the table as the City of Austin revamped how its police cadets are trained.

She said she takes issue with the modeling in the APD staffing report.

“The entire program is based on the idea that we are never going to use our officers differently than we use them during the period in which this data was created,” Mitchell said. “In fact, we have already started using our officer time differently.”

During a media briefing Tuesday, APD Chief Joseph Chacon said the report would be only one tool used when determining staffing. Chacon is expected to present a detailed plan to the city council in the coming weeks.

PD staffing issues not unique to Austin

The Austin Police Department is not alone when it comes to staffing challenges. While hiring slowed slightly last year nationwide, the real damage came from resignations and retirements.

That is according to a special report from the Police Executive Research Forum which surveyed 194 police departments across the country.

Responding agencies reported an 18 percent jump in the resignation rate in 2020-21, compared to 2019-20.

Retirements were even more widespread. The trend in all of the departments surveyed showed an increase of 45 percent.

It is worth mentioning that there are more than 18,000 police departments across the country, so the 194 responding agencies in the survey is a fairly small sample size.

Copyright 2022 Nexstar Media Inc. All rights reserved. This material may not be published, broadcast, rewritten, or redistributed.

Austin-Travis County


More reviews

Tracking the Coronavirus

Coronavirus Cases Tracker

Latest Central Texas COVID-19 Cases

Trending Stories

Don't Miss