Analysis is the process of pulling meaning from chaos. When you first load up your agency in Community Crime Map, you may feel overwhelmed by all those dots on the map. What does it all mean? Is crime high in my area? Does the type of crime happening near my home pose a danger? When is all this crime happening? This blog post will teach you some basic crime analysis techniques so you can make sense of crime in your area. Empowered by this knowledge, you can help make your community a safer place.
Choose a specific problem you want to address and define it clearly as either a question or a statement. For instance, you can state your problem as an exploratory question such as What time of day do burglaries most often occur within a mile of my home?. On the other hand, you can describe an assumption you hold that you wish to test such as Burglaries occur most often at night. We’ll use these two problem definitions as our examples for the rest of the post.
Use the address buffer, crime type list and date range selector to view only the crime that is relevant to your specific question. Follow the steps below to filter out all the unnecessary data so you’re left with only the crime you care about. Feel free to follow along with your own address.
Now you are viewing only the crime relevant to your analysis. Feel free to check out the Data Grid tab to get a feel for the nature of the crimes. You might also want to take a look at the Metadata tab to see the quality of the data, which has a direct bearing on the quality of your analysis.
The next step is to select the appropriate analytics for our problem. We’re trying to discover patterns dealing with the time of day. This is a temporal problem. Consequently, the most appropriate analytic for our problem is the Temporal Topology, which displays when crime is occurring in time. The x-axis represents the hour of the day while the y-axis represents the day of the week. Times with the most crime appear in red while times with no crime appear in blue.
To run the temporal topology analytic, simply click on the Analytics tab.
Use the analytic you ran in the previous step to draw conclusions about the specific problem you defined in step one. Let’s start with our first definition, which we posed as an exploratory question: What time of day do burglaries most often occur within a mile of my home? Look at the temporal topology graph and answer that question by looking for hotspots (areas with the most crime) as depicted by red, orange and yellow shades. What times during the week show the most activity? In our example (shown above), it appears that burglaries occur most often during the day with slightly more activity during business hours.
Ask the opposite question: What times have little to no burglaries? In our example, the early morning hours (2 a.m. to 7 a.m.) of the weekdays show very few burglaries.
You can flesh out this analysis a little bit by looking at which days have more crime in general. Confirm this by looking at the Crime Type by Day of Week graph. Likewise, take a look at the hours of day that tend to have a good deal of crime no matter the day of the week.
In our example, the number of burglaries don’t vary too much between the days of the week. Although we do notice a slight dip in the middle of the week, it only differs from the other days by one or two burglaries; we would need to include much more data to draw any solid conclusions from the Burglary by Day of Week graph.
Let’s answer our second problem definition, which we stated as an assumption that we wanted to test: Burglaries occur most often at night. We have probably already answered this question in the above analysis, but look at the temporal topology graph again with this assumption specifically in mind. Do burglaries happen more often during the night? Or do they tend to occur more often during the day?
What are some ways we can improve this analysis? Let’s hear your thoughts in the comments. Hint: Does our problem definition say anything about the distance from our address?
Now that you have analyzed a specific crime problem in your area, don’t let that knowledge go to waste. Here are some steps you can take to make your community a safer place with your analysis.
Those are just a few ideas to get you started thinking about where to go from here. Suggest more action steps in the comments. We might even add them to the list! We would also love to hear how you are using Community Crime Map to analyze crime in your area. Your experiences will help other readers become better analysts, so post away!
Make sure your problem definition is specific. Too broad of a question will not exclude enough of the chaos to glean real meaning from the data. Analyze one crime type at a time instead of many. Focus your analysis on a small area rather than a large one. Compare two crime types over a period using the Timeline graph of time instead of trying to find patterns in all the crime types at once.
Crime, as with anything else, can randomly spike well above the average rate one week and dip real low another week. As a result, try to include as much data for your specific crime types or time periods as you can to even out these ups and downs. This will give you a more accurate picture of the real crime rates. For instance, when testing our assumption about when burglary most often happens, we should turn off the address buffer and bring in several months of data.
Crime analysts at your police department have years of training and experience as well as powerful analysis tools. So you may want to check with them to confirm the results of your analysis and learn what you and your neighborhood can do to address the problem.
Density Maps or “Hotspots” are a huge source of fascination for the public and law enforcement. The attraction is obvious—a picture is worth a thousand words—however, the science behind actually identifying a hotspot varies widely from user to user. Analysts are always looking for clusters, groups, and hotspots. Indeed, trying to use a set of points—such as crime events—to define an area such as a hotspot, hunting ground, activity space, etc., is an important activity for any mapper. After all, the police need to narrow their search, prioritize their deployments, or focus on an area when at all possible.
By analyzing the density of points, rather than their mere locations, it is possible for us to visualize the influence of events very clearly and to convert the locations of discrete points into areas of interest. These in turn can give us insight into where future events may occur, from where they may originate, and why certain targets may be selected.
Density is calculated by counting up the number of events within a selected range of each cell; cells with a higher count of nearby events have a higher density than cells with a lower count. This range, known as the Search Radius, must be chosen carefully. If the Search Radius is short, there may not be any cells which are in range of more than one or two events. On the other hand, if the radius is too large, every cell might be in range of every event, therefore giving a meaningless result. Search Radius selection is the most critical part of performing density analysis. RAIDS has a proprietary algorithm that determines a suitable search distance at any zoom level in hopes of providing you hotspots at any level.
Imagine the topology as being like a map, except instead of terrain, it maps time. The X-Coordinate (horizontal) indicates the hour of the day, while the Y-Coordinate (vertical) indicates the day of the week. The Z-Coordinate (elevation) is represented by the volume of activity at that hour and day.
Temporal topologies identify the level of crime activity at a particular day and hour.
You read the temporal topology just as you would read the density map in RAIDS. Look for areas of high (or low) activity across the 168 hour week. Police agencies use the temporal topology to understand when activity happens, when to deploy resources, and when to staff officers. Temporal topologies, using vivid colors and shapes to draw the reader’s attention to significant findings, can be much clearer and easier to interpret than reading through countless rows of mind-numbing cross tabulated data