Visualizing Calls for Service (CFS) Data
This blog post was originally written in 2020, but the concepts are still relevant today.
Calls for Service (CFS) Data is considered one of the best sources of information to inform police departments about calls from the public and what police are doing day to day and hour to hour. This post will show how Microsoft Power Bi can be used to pull open-source data off the web and create a dashboard that is informative to policymakers and the public.
In policing, one of the best sources for understanding what is happening in the community is Calls for Service (CFS) data. When someone calls the police and requests law enforcement, a “call for service” is generated by the call taker. Captured are a call type (what the situation is), date, time, location, caller information, and potentially many other points. This call for service is then dispatched to an officer (or officers depending on the call type) and the officer responds.
Out of all datasets used in the profession, calls for service are often considered the purest form of data. This is because it captures exactly what is reported by the community member when they identify a problem and request the police for help.
Once the officer gets to the call, the officer may or may not take a report on the incident. Most of the time, the officer has a great influence on whether a report is completed or not. Thus, official reports are dependent on the officer making the effort to document the incident. Calls for service information, therefore, have no officer influence and are considered to be more reliable of what is actually occurring in the community.
It is a matter of fact that many officers will work hard to not generate an incident report. Thus, some officers will convince a victim of a crime to not pursue an official report and the incident is never officially recorded. This is problematic, but even the most stringent policies and supervision will not avoid this. Of course, the most serious of crimes will generate reports, but it is the smaller crimes- the quality of life issues, the disorder calls, etc - that have major consequences on larger crimes within a certain area. That is why calls for service is preferred when understanding what is going on in a community.
Furthermore, calls for service data - specifically reactive calls - is the best dataset to use to ensure strategies are not informed by biased data. There has been a lot of talk lately about how police data is inherently biased against minorities (a topic that will be explored in a later post). Reactive CFS, though, are only those data points where a community member called the police. This data is simply, what community members are reporting. (Officer Initiated Calls are a different category and are explored below).
With that said, if police-community relations are so bad that the police are not contacted when there is a problem, then the data will be flawed. However, most CFS is reflective of actual incidents in the field and is one of the best sources available for police departments.
The blog published by RTI International, which I reference in a previous post, identifies the need for policing to leverage CFS data to help inform police on a day-to-day basis.
As an example of how CFS data can inform police, I used Seattle PD’s open data to create the dashboard above in Microsoft Power Bi:
The dashboard contains two pages, one reflecting Reactive Calls for Service, and the other showing Proactive “Officer-Initiated Activity”. It is vital to separate the two kinds of CFS data because they are quite different and provide different insights.
Reactive CFS data is what the community is reporting to the police. Thus, the data consumer can get a grasp of exactly what is of concern to community members. On this dashboard, the overall reactive CFS trends by week of the year are visualized using a line chart. This provides a quick overview of whether the agency is more or less busy. The table below the line chart is sorted by most CFS to least for each type of call. This is important to understand which types of calls officer are being sent to the most. To the left of the table is the average response time for calls for service. By clicking on individual call types, the viewer can see how long the average response is to that type of call. In this data, it is evident that calls are triaged by priority and in-progress violent crime calls have a much shorter response time than past-tense call types. Lastly, the beat map on the left side of the dashboard shows where most of the calls occur. Clicking on an individual beat will adjust the table to identify what types of calls are most prevalent in that area.
The proactive CFS dashboard is remarkably interesting. It is immediately apparent that proactivity by offices has declined significantly since week 22, which coincides with the significant unrest in Seattle due to protests of police brutality. Seattle has been in the news almost daily due to the issues the city has been facing due to unrest. The proactivity by police officers is clearly reflected here.
Similar to the reactive dashboard, this page shows the types of proactive work that are logged by officers as well as a map of what type of work is going on in each beat. A Key Performance Indicator (KPI) shows an example of goal setting and whether it is being met. Police leaders may want to track a standard of proactive work being completed and the KPI offers a quick look at whether that goal is being met or not.
Data dashboards offer a lot of information to police departments from the patrol officer to the chief. They also can provide transparency for the community to get a better understanding of what their public servants do. With so much discussion about the role of policing in society, exploring and understanding this data can inform conversations and lead to meaningful change.
Want to learn more? Head over to the "Contact" page and leave me a message. I'd love to discuss this more!
Commentaires