Crimestat is a spatial statistics program that analyzes the location of criminal incidents.

This is the type of software used by police, military intelligence and marketers. It measures the frequency and probability of events from a set of data. Why software? Because doing the math by hand then mapping it out is extraordinarily painful and time consuming. Computers have perfect memory and perfect calculation.

Here’s a good summary of the software. Crimestat was originally used to map and analyze crime data by the Baltimore Police Department.

Crimestat uses an XYZ grid, for longitude, latitude, and intensity/ID. This maps out locations of criminal events onto an urban map.

If you graph it out, you create a topographical map. The points of interest are located on the XY grid, and their intensity measures their “height” or weight. With this, analysts can identify patterns, clusters, and hot spots. To visualize it, it looks like a “hill” on the city map. That’s your crime-ridden neighborhood. The software also calculates the travel routes used by criminals, as seen here.

The software gives more exact data, including simple measurements such as the maximum, minimum and mean distance between points, and the relative intensity of events.

Most crime and other patterns of behavior follow Normal Distributions. Police can measure the mean, mode, and median crimes and standard deviation to predict the probability of crimes in regions of the map. This is how they properly identify trouble spots and place more forces in those regions to lower crime rates.

Crimestat compares the mapping of points to a random distribution. It runs Monte Carlo random simulations then compares the random data to the real data. This detects non-random clustering compared to random clustering and raises the confidence level that these clusters are not coincidence.

Crimestat has a number of features.
HotSpot Analysis calculates the frequency of incidents in each neighborhood by mean, mode, max/min. It can calculate the Neighbor Hierarchical Spatial Clustering and other patterns.

It also does space-time analysis as crime rates change over time.

Crimestat also measures spatial distances like traveling time and journey to crimes. For instance, on foot, a person moves 5 miles per hour. In a car on a highway, he moves 60miles per hour. You can measure type of traveling to and from a crime scene to track criminal movements. This maps out the travel networks used by criminals.

The Correlated Walk Analysis studies the sequence over time of crimes committed by one criminal or organization. It crunches data describing the time, distance, and bearing between events. This narrows the range of distance where the criminal operates. CWA can measure, within a Gaussian probability distribution, the range of locations and times where the criminal struck. With historical data, the police know the captured criminal’s point of origin. Crimestat compares this with the journey-to-crime data and the CWA data to describe criminal movements and behavior. It can literally draw out the trips to and from events over space/time.

With good data on criminal movements, CWA can also help track the criminal’s likely point of origin. Again, this is based on probability, not magic. All human beings are creatures of habit and they do not deviate that much from one another. A dataset of past criminal behavior can help estimate probable behavior of present criminals.