Traffic Safety Camera Analysis
To assess the effectiveness of traffic safety cameras installed between 2008 and 2012, the Office of Performance and Accountability conducted an analysis to measure whether camera locations experienced fewer crashes on average than comparison sites after installation. Instead of manually choosing the set of comparison locations, OPA used a machine learning approach to match each camera installation to a control group with similar safety profiles before installation. After generating the comparison groups, various statistical tests were carried out to determine whether the camera locations differed significantly from their control sites.
In line with a literature review prepared by the National Transportation Safety Board, the analysis found evidence that traffic cameras installed in New Orleans from 2008 to 2012 did have a beneficial effect in terms of reducing crashes at the camera locations. While there was considerable variation among camera sites, road segments with cameras had approximately 21 percent fewer crashes on average than would have otherwise been expected, even after controlling for confounding factors. In addition, 76 percent of camera locations experienced a smaller increase in crashes than a set of matched comparison sites, with an average difference of 23 percentage points. Similarly, 54 percent of camera locations had a decrease in the crash rate, compared to 9 percent for comparison sites.
Because significant variance existed among locations, however, OPA cautions that estimates of average effect size should generally be interpreted as directionally correct, rather than numerically precise.
EMS Ambulance Posting Location Optimization
In an effort to promote faster, more reliable, and more equitable response times to health-related 911 calls in the face of demand for its services, New Orleans Emergency Medical Services (NOEMS) worked with OPA to develop a new, data-driven ambulance placement protocol. In partnership with LSU analytics graduate students, OPA developed an optimization algorithm that ranked locations by potential impact, incorporating daily and weekly patterns in traffic, 911 calls, and ambulance availability. The result was a prioritized list of locations to place ambulances during day shifts (assuming high traffic and daytime 911 call patterns) and the night shifts (assuming low traffic and nighttime 911 call patterns, which concentrate in the French Quarter and Central Business District).
Analysis of 911 data shows that since the new protocol was implemented, there are more equitable response times across districts, with the largest gains (20%) in the proportion of calls within response times standards (12 minutes) coming in the 4th district (West Bank), the area of the city with the historically slowest response times. Response time compliance in the 7th district (New Orleans East), a district with historically slower response times, also improved substantially (9%). In addition, the new protocol led to modest, statistically significant improvements in citywide response times during the night shifts. During the day shifts, there were no overall statistically significant improvements in response times. However, equity in response times across districts improved.
Targeted Smoke Alarm Outreach Program
As a component of its fire prevention effort, the New Orleans Fire Department (NOFD) offers free smoke alarm installation to all city residents. The initiative includes door-to-door outreach where firefighters will go throughout the city and install alarms for those in need. To ensure that the campaign has the greatest effect, NOFD worked with OPA to design a model that prioritizes areas of the city that are most likely to suffer fire fatalities and least likely to have a smoke alarm.
Code Enforcement Abatement Tool
In order to take action on blighted properties, decision makers in Code Enforcement must be able to determine if a property should be demolished or proceed to a code lien foreclosure sales (Sheriff’s sale) where it can be sold to a third party for redevelopment. In order to make abatement decisions faster, more consistent, and more rigorous, OPA worked with Code Enforcement leadership to develop a decision tool utilizing methods employed by other machine learning recommender tools. Enigma provided valuable support on the project.
Primary Care Text Campaign
The Greater New Orleans Community Health Connection (GNOCHC) provides health care coverage to more than 60,000 low-income people in the New Orleans region. While GNOCHC enrollees are entitled to free, annual appointments with their primary care doctor, only about half had taken advantage of this over the previous two years as of 2014. This number is especially concerning because of New Orleans’ poor health outcomes: 12% of babies are low birth weight, compared to 8% nationally, 12% of residents are diabetic, compared to 9% nationally, and New Orleans is consistently in the top five among US cities for HIV, and syphilis case rates. Since many of these cases can be address, prevented or mitigated in a primary care setting, getting people into care is more important than ever. To encourage GNOCHC enrollees to set up an appointment with their primary care physician, OPA worked with the City’s Health Department, the City’s Office of Neighborhood Engagement, and 504HealthNet on a primary care text campaign. Behavioral Insights provided valuable support on the project.
For those interested in exploring further:
- The Harvard Kennediy School maintains a website on using data in cities.
- The University of Chicago Data Science for Social Good website provides examples of their own work in cities