Google Street View, machine learning, and social patterns

I have wondered why more researchers do not make use of Google Street View. Here is a new study that connects vehicles in neighborhoods with voting patterns and demographics:

Abstract: The United States spends more than $250 million each year on the American Community Survey (ACS), a labor-intensive door-to-door study that measures statistics relating to race, gender, education, occupation, unemployment, and other demographic factors. Although a comprehensive source of data, the lag between demographic changes and their appearance in the ACS can exceed several years. As digital imagery becomes ubiquitous and machine vision techniques improve, automated data analysis may become an increasingly practical supplement to the ACS. Here, we present a method that estimates socioeconomic characteristics of regions spanning 200 US cities by using 50 million images of street scenes gathered with Google Street View cars. Using deep learning-based computer vision techniques, we determined the make, model, and year of all motor vehicles encountered in particular neighborhoods. Data from this census of motor vehicles, which enumerated 22 million automobiles in total (8% of all automobiles in the United States), were used to accurately estimate income, race, education, and voting patterns at the zip code and precinct level. (The average US precinct contains 1,000 people.) The resulting associations are surprisingly simple and powerful. For instance, if the number of sedans encountered during a drive through a city is higher than the number of pickup trucks, the city is likely to vote for a Democrat during the next presidential election (88% chance); otherwise, it is likely to vote Republican (82%). Our results suggest that automated systems for monitoring demographics may effectively complement labor-intensive approaches, with the potential to measure demographics with fine spatial resolution, in close to real time.

And a little more explanation from a news source:

The researchers created an algorithm to identify the brand, model and year of every car sold in the US since 1990.

The types of cars also provided information about the race, income and education levels of a neighborhood, the study said.

Volkswagens and Aston Martins were associated with white neighborhoods while Chryslers, Buicks and Oldsmobiles tended to appear in African-American neighborhoods, the study found.

This study seems to do two things that get at different areas of research:

  1. Linking lifestyle choices to voting behavior as well as other social traits. Researchers and marketers have done this for decades. For example, see this earlier post about media consumption and voting behavior. This hints at the work Bourdieu who suggested class status is defined by cultural tastes and lifestyles in addition to access to resources and power.
  2. Connecting different publicly available big data sets to find connections. Google Street View is available to all and election outcomes are also accessible. All it takes is a method to put these two things together. Here, it was a machine learning algorithm by which different kinds of vehicles could be identified. It would take humans a long time to connect these pieces of data but algorithms, once they correctly are identifying vehicles, can do this very quickly.

Of course, this still leaves us with questions about what to do with it all. The authors seem interested in helping facilitate more efficient national data-gathering efforts. The American Community Survey and the Dicennial Census are both costly efforts. Could machine learning help reduce the effort needed while providing accurate results? At the same time, it is less clear regarding the causal mechanisms behind these findings: do people buy pick-up trucks because they are Republican? How does this choice of a vehicle fit with a larger constellation of behaviors and beliefs? If someone wanted to change voting patterns, could encouraging the purchase of more pick-up trucks or sedans actually change voting patterns (or are these more of correlations)?

Using Google Street View to collect large-scale neighborhood data

One sociologist has plans to use a new Google Street View app to study neighborhoods:

Michael Bader, a professor of sociology at the American University, revealed the app developed is called Computer Assisted Neighborhood Visual Assessment System (CANVAS). The app rated 150 dissimilar features of neighborhoods in some main metropolitan cities in the U.S. The researchers claim the latest app reduces the cost and time in research.

With the help of Google Street View, the new app connects images and creates panoramic views of the required rural areas as well as cities. Bader explains that without the Google app researchers would have to cover many square miles for data collection, which is a painstaking job…

The app has already received funding of around $250,000 and s also supposed to be the first app that examines the scope and reliability of Google Street View when it comes to rating the neighborhoods in the U.S.

Bader reveals he is currently using CANVAS for a research on the Washington D.C. area. He revealed the population of people who have reached 65 and over in the region will be 15.3 percent by 2030. Bader hopes to understand why elderly people leave their community and what stops them from spending the remainder of the lives in the region. Bader’s research wants to understand the challenges elders face in Washington D.C.

As an urban sociologist, I think this has a lot of potential: Google Street View has an incredible amount of data and offers a lot of potential for visual sociology. While tradition in urban sociology might involve long-term study of a neighborhood (or perhaps a lot of walking within a single city), this offers a way to easily compare street scenes within and across cities.

Viewing the insides of stores on Google Maps

Adding to its Street View capabilities, Google also will allow browsers to see inside some retail establishments that allowed Google to photograph their interiors:

A test program launched in April of last year was bearing fruit in a growing array of panoramic images taken inside businesses that volunteered to be part of the project.

“We’ve been seeing renewed interest in the past few days because, as promised, we’re getting more imagery online,” Google spokeswoman Deanna Yick told AFP on Monday…

Small businesses in Japan, Australia, New Zealand, and the United States have been able to invite Street View photographers into their shops or eateries to capture images then served up with Google online maps.

“With this immersive imagery, potential customers can easily imagine themselves at the business and decide if they want to visit in person,” Google Maps product manager Gadi Royz said in a blog post early this year.

My big question: will this actually bring more customers inside the shops? I’m skeptical: how many times would someone be wondering about whether they should visit a store, look up the interior image on Street View, and then make a positive decision. What if the image is actually a negative thing, perhaps due to the lighting (I wonder if they adjusted for this), outdated decor, or, for lack of a better term, a lack of “coolness”?

We could also ask whether Google’s efforts in these areas actually encourage in-person community. If given more information in general through search engines, images, and reviews (with Google recently buying Zagat), will people be more likely to venture out of their homes or away from their internet-enabled devices? Will they become overwhelmed with the choices (like Barry Schwartz argues in The Paradox of Choice) and be less likely to choose any?

In the end, Google must think that providing these interior images are going to help them make money.

With all of those cameras watching, few people change their public behavior

A Canadian sociologist argues that although more people are being watched in public, through phenomenon like Google Street View and a multitude of security cameras, few people are changing how they act in public settings:

Nathan Young, an assistant professor at the University of Ottawa’s Department of Sociology and Anthropology, believes few Canadians have altered their public behaviour.

“When we’re out on the street, there’s an understanding we’re in public and there’s a risk of being seen or talked to,” says Young. “What’s different now is if we go outside to change a tire in our underwear, it can be exposed worldwide.”…

Voyeurism has always been a part of human motivation, says Young, an authority on privacy issues.

But ogling the Google images also tells us something about our sense of fun, he explains, pointing out some of the images are theatrics or misunderstandings of a moment.

“Clearly, there are people out there that want to play (with the technology),” he reasons. “Not for politics or protest. Just a personal imprint.”

Young says there’s no evidence people are more aware or cautious as they head out their doors.

At this point, if you are acting “normally” in public, you don’t have much reason to worry that some camera out there might see what you are doing. At the same time, there seems to be a decent number of people who are worried that these cameras and technologies could end up being used against them.

It is interesting to note how people do act in public and to think whether this differs from how they act when they are alone.