Using Data to Prioritize Demolitions

Detroit faces a huge problem with vacant housing units. The Census’ American Community Survey estimated there are 360,951 housing units in Detroit. Of these, it estimated that 253,629 are occupied, indicating that 107,322 are vacant. Given sampling error, these estimates could be off by as much as 10,000 units at a 95 percent confidence level. Even so, that means an estimate of nearly a 100,000 or 29 percent of the city’s housing units could be vacant. Many of these units are vacant and open, and not a small number have burned.

Each vacant and open unit acts like a black hole on the economic values of homes in Detroit, dragging down the values of the neighboring homes. And these become locations for illegal activities as well as environmental hazards. So, there is an urgency to demolish, preserve or repair as many as possible. However, a cash-strapped city can afford to do only so much, and Mayor Bing has said he would like to demolish 10,000.

Even so, there are many thousands of additional structures in Detroit that pose threats to city residents; the city does not currently have the funding to demolish all of these structures though. Because of this, decisions must be made about which structures to target first.  Dr. David Martin and Dr. Lyke Thompson at the Wayne State Center for Urban Studies have begun  working on a point system to identify priorities for demolition for the structures identified by the city. In order to identify the most dangerous buildings, this method considers for each address nearby crime reports, past reports of lead poisoning at that address, and the proximity to schools.

To make this work clear, we first present some overall maps of the city and then drill down on one neighborhood—the Osborn area, located in Northeast Detroit(see the map of Detroit below). Osborn, as defined for these purposes, includes many healthy areas and some areas which have experienced high rates of foreclosure and vacancy. There are many efforts underway to substantially improve the area, so it would especially valuable to identify the housing units that need to be demolished first. This is an initial effort to concentrate on housing that may produce the greatest externalities. We fully recognize that with more data this approach could be improved.

Slide04

Slide05

While the Osborn neighborhood is explicitly examined in this post, this map shows the frequency of lead poisoning incidences per address/home in the City of Detroit as whole. The yellow dots show there are 4,610 homes in the city with two occurrences per home; these are the most frequent. However, the blue dots, which show  three to five or six to nine occurrences per home, cover more of the map because of the higher number of people affected. There was one home in the southeast portion of the city that had 17 lead poisoning cases, according to the map.

map

The above map shows where all K-12 schools in the City of Detroit are located. While this post only examines the Osborn neighborhood, this map provides readers with a better understanding on the number of schools in the city and  how the point system can be used beyond the example in this post.

Slide07

The above map displays the hot spots for crime, which includes property crime and violent crime, in the Osborn neighborhood of Detroit. Red means that is the “hottest” spot for crime in the area, orange/yellow comes in second, followed by green and purple. If there is no color that means it is a “cool” spot, with next to little or no crime. The area with the highest amount of crime is near Seven Mile and Gratiot. When performing hot spot mapping it should be noted that the hot spots are relative to the population size of the area being examined.

Slide09

The map above shows the locations of all structures currently on the demolition list for the Osborn neighborhood. Some structures have already been slated for demolition, through an initiative from Detroit Mayor Dave Bing’s office; these are shown with yellow icons on the map.  The red icons represent 820 structures that need to be demolished according to the WSU/CUS point system but have not been funded or scheduled, according to city data. There is a high concentration of buildings near Seven Mile Road between Redmond and Schoenherr. There are also several clusters of addresses in and around the southern area of Osborn, east of Hoover Street.

As can be seen below, the number of buildings in need of demolition with no funding is greater than the number of homes slated to be demolished. So how does one establish priority? One way is a point system.

Slide11

Our first effort to create a point system relies on data that the Center has available—crime data, lead poisoning data and distances to a school. Each unit on the list of unfunded structures for demolition was given a score, where:

•Ten points were given to each address within a quarter mile of a school;
•One point was given for every crime reported on the block where the structure was located;
•Two points were given for each instant that a different (as opposed to the same) child was identified as lead poisoned at an address, between 1988 to 2012.

In the map above, the each icon represents a prioritized structure with the larger the icon representing a higher score. According to this map there is a high concentration of homes near Seven Mile Road between Redmond and Schoenherr streets where units were calculated to have a high priority for demolition.  There are also several clusters of addresses in and around the southern area of the Osborn neighborhood, east of Hoover Street.

It is possible to add factors to this analysis or to adapt the weighting of particular factors. For example, vacant housing with a large amount of combustible vegetation close to housing adjacent units would create a risk of spreading fires.

In any case, unless massive funding for demolition becomes available, priorities could help cash challenged Detroit focus on the worst units first.

Advertisements

One Response to “Using Data to Prioritize Demolitions”

  1. Alex B. Hill Says:

    This is very comprehensive information and use of data. Would you be willing to share the datasets used? Thanks!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s


%d bloggers like this: