So let’s start piling in the data! Right?
Not just yet cowboy; pull back the reigns for just a minute. I know you’re anxious to get your Intelligence Led Policing initiative up and running, but don’t skip the most important step: making sure the data you are going to feed into your intelligence system isn’t garbage! We’ve all heard the term, “garbage in, garbage out.” Well, the data integration between an RMS/CAD system with an analytics solution is where that term really comes to life. So many agencies I speak to have unwittingly made the mistake of pushing data into an intelligence system without vetting that data first. What makes it worse is that most of these agencies won’t figure out that they pushed bad data into their system until they begin to get results from that system that doesn’t make sense. Some of the most common results they will see as a result of bad data are:
- Maps will be a mess
- Multiple crime types getting bulked together (think murder, rape, theft, lumped together in an “other” category)
- And generally, their numbers won’t jive with what they know to be true
So let’s talk about your data; I mean really talk about why it is so important to spend time on your existing data set to make sure that it is clean, standardized, and accurate. Don’t assume that your data set is correct. It may be a difficult thing for you to do, but I want you to assume that you data set is “generally” correct, but needs verification before it is trusted. As President Reagan said so often, “Trust, but verify”. Now don’t get me wrong, you can certainly use your existing data, as it sits, without verification, for your intelligence initiative. You can pull intelligence from that un-vetted data set and distribute that intelligence throughout your department. But, without verifying your data, your intelligence will be wrong.
Data verification is the key to good intelligence.
So how do you verify your data? Here are a few areas to check in what I would consider the most common error categories, and steps to take to correct any mistakes you may find.
1. Maps - Most commonly our mapping data, whether from GIS, Google, Lat/Long, or address, is far from perfect. As a matter of my experience, I would say that maps are one of the most inaccurate, yet one of the most desired data sets for an agency.
Crime map data comes from a variety of technologies.
Meaning, we all want good maps, but very few of us actually have such.
Why? – There are a multitude of reasons that our maps leave a lot to be desired, such as: inaccurate input from our officers and dispatch, duplicate addresses resulting from some unknown oddity in city planning, and GIS, Google and other mapping systems simply putting the map point in the wrong place. Very few of these, other than officer/dispatch mistake, are under our control. This is what makes the mapping issue such a big problem, many of us just throw up our hands and exclaim, “it’s out of my control, so we’ll just have to live with it.”
Effect – By leaving bad mapping to its own devices, we are allowing our data to be tainted. In essence, we are saying, we don’t really care about those crimes that are mapped incorrectly, we will just rely on the crimes that are mapped correctly. Of course, this is false logic because, as we all know, especially Intelligence Led Policing should be an “all crimes” approach. With an “all crimes” approach, we know we are getting the entire picture. Without it, we are only getting a biased view.
Solution – The first step is training – make sure your officers and dispatchers are entering the addresses correctly. After that is done, choose a software solution, such as the one I use, CommandCentral by PublicEngines, that allows you to identify incorrect mapping points and move them to their correct location. I have seen many software solutions, records management systems included, that allow the user to see incorrect mapping points, but very few allow the user to move those bad points to where they really need to be. CommandCentral, however, has gone one step further than just allowing you to identify and move your mapping points, it has streamlined that entire process down to just a few clicks of the mouse. I am able to look at all of my crimes over a particular date range in the entire city, un-click my zone designations, and then click a tab called “outside” that shows all of my crimes that mapped outside of my zone designations. I can utilize the administration tools to move mis-mapped crimes to their proper location. A quick solution to a menacing problem.
2. Crime Types – Used widely by agencies to easily designate the difference between certain crimes, such as Burglary Residential and Burglary Commercial, these designations have quickly gotten out of hand in many jurisdictions. What I mean by out of hand is that many agencies have found themselves with so many crime types, some have 400, 500, or more, that they have a hard time keeping the designations separate.
Why? – Let’s face it, so many times it is much easier for an officer to pick Theft/Other than it is to find the exact designation that the crime demands, especially when there are so many options. Whats more is that in so many records management systems, in order to run a report on a parent crime type such as theft, you have to run multiple individual reports on each crime type designation within that parent theft designation. And sometimes, an officer incorrectly categorizes a crime. He may write a burglary report, when in actuality, it should have been a theft report.
Effect – You find your data is scattered all over your records management system. For instance, in order to run a report of the thefts in your jurisdiction, you have to go to multiple report tables, run multiple reports, and then you still have to compile, generally manually, all of those records into one document.
Solution – Like Maps, the first step is training. Make sure you train your staff the proper way to designate each crime and more than that, make sure they know why it is so important to properly designate each crime. The second step then is to revert back to your software solution to make this whole process easier for you. I again refer to CommandCentral, it allows me to easily bulk my various crime types into easy-to-understand quick tabs. Meaning then, that all of my various thefts are bulked into one tab simply called “Theft,” while still allowing me to just choose one type of theft and generate a report on it alone if needed. The system also allows me to bulk these crime types myself. That way, if I want to move a specific crime type to another category, I can do so with ease.
3. Numbers – Truly at the heart of the matter aren’t they? The chief wants them, the city council wants them, even the FBI wants them. But if the data problems we spoke of earlier are not right, then your numbers are surely off as well.
Why? – There is a direct correlation between your numbers being off and the data supporting your numbers being off. As go the earlier topics, so goes your numbers, and so on.
Effect – So many times we are in a rush to get the numbers out that we forget the correlation. As a result of putting out bad numbers, we all look poorly to those we have created the report for. We are working with something the likes of a living organism, all parts must work together for all the parts to work correctly.
Solution – Begin by correcting all of the areas that you would get your numbers from, such as the areas we spoke of earlier. Please be diligent about this. You must understand that if your data sets are not correct, then there is no way your numbers, which come from those data sets, can be correct.
In closing, I want us to all remember an old saying some of us were taught in mandate school: “The Fruits of the Poisonous Tree.” Use it as a guide for our Intelligence Led Policing. If we use bad data (the poisonous tree) in our intelligence initiative, then the intelligence that we get out of that bad data (the fruit), is corrupt, misleading, and all around garbage. Your Intelligence Led Policing initiative lives and dies on the quality of the data you feed it. Feed it good, accurate data, and it will thrive, feed it fruits of the poisonous tree and it will wither and die.