GIS

[FEMA Camp Administrator]

Anyone here familiar with using geographic information systems software? Would you have any recommendations on free applications, and/or those that are easy to use? I'm currently running a heatmap program in QGIS based directly on an online tutorial and am trying to get the hang of it, but it's far from intuitive, and either this isn't an exact science, or it's a science well beyond my comprehension.

My immediate (end of semester) goal would be to simply be able to produce a series of crimes on a municipal map and measure distance from one another through a heatmap/kernel density application. I don't know if I'd be able to measure time as well, or if I'd have to do that through Excel.

[cinyc]

I use QGIS for all my election maps. Jimrtex and I put tutorial on this site about this time last year:
https://uselectionatlas.org/FORUM/index.php?topic=276225.0

I also have a bit more of a tutorial as my pinned Tweet on Twitter (@cinyc9).

QGIS has since upgraded from version 2 to 3. Most of the stuff from the tutorials is still the same, though they renamed things (Print Composer is now Print Layouts, for example) and changed some of the menu hierarchy.

If you have some questions, let me know. I've never tried to measure distance or time, but can probably figure it out.

[jimrtex]

Anyone here familiar with using geographic information systems software? Would you have any recommendations on free applications, and/or those that are easy to use? I'm currently running a heatmap program in QGIS based directly on an online tutorial and am trying to get the hang of it, but it's far from intuitive, and either this isn't an exact science, or it's a science well beyond my comprehension.

My immediate (end of semester) goal would be to simply be able to produce a series of crimes on a municipal map and measure distance from one another through a heatmap/kernel density application. I don't know if I'd be able to measure time as well, or if I'd have to do that through Excel.

A feature of QGIS is that it is very extensible and there is a large community contributing extensions.

You can usually find a lot of information through a Google search. I searched for:

"QGIS heatmap plugin"

And there is a plugin for that.

I don't know what you mean by time in this application.

What is the format of your crime data?

[FEMA Camp Administrator]

Hey gang, thanks for responding.

Jmrtex, I've been able to get on QGIS as far as the Heatmap application and was able to produce one through a tutorial, though the data handed me by my supervisor is either tainted or I'm in a situation of criminal incomprehension of how to manipulate this data.

As for my project, the "theory of near repeat victimization" essentially states that people/buildings/organizations near victims face temporarily elevated risk. Research projects on near repeat victimization tend to measure it in series of hundreds of meters/feet (for example, a chart showing the amount of subsequent burglaries occurring within 0-100 meters, 101-200 meters, etc.) and days/weeks. A composite chart in a study might show elevated risk within a series of categories distinguished by spatial/temporal distance from an initial offense. I can attach a few papers if you're curious. In theory I can somehow jigger a measure of time on an excel sheet; there are a few methods, including GIS, that I might be able to use to measure space. Right now I'm just trying to get a grasp of it so I can manipulate it to suit my purposes in the future. It's also an invaluable skill in certain fields.

My supervisor has yet to send me the actual burglary information.

[jimrtex]

So you have x,y,t values representing reported crimes. Are there other attributes that the data might be filtered on?

Depending on the format of the data it might be simplest to use Excel to get the data into a CSV file and input that into QGIS.

You might be able to do a 3D display of x,y,t assuming you can find a suitable scaling for the t-values but that would be hard to comprehend, but you might be able to see spatial/temporal clusters.

You could bin the x.y values into a grid, and produce a 2D histogram, but that might not be difficult to find the correct grid size. So that leads to a heat map which is a continuous density function. This will show where crime is concentrated, but won't be time dependent.

Can you compute a 3D heatmap?  If so, you could time slice it and put it into a video such as cinyc's animated GIF's. If the theory or near repeat victimization is correct you should see something that looks like a radar precipitation map for scattered showers, with a storm popping up in green as the first crime in an area occurs, and then intensification toward red and then turning back green as the thunderstorm/crime spree passes.

If you can't do a 3D heatmap, you could still filter the data by time intervals and produce 2D heatmaps.

Or you could take all pairwise combinations of events and calculate the delta d and delta t, and then generate a heatmap for that data. This won't show spatial location of crimes or when they occurred, but will directly show the relationship between distance interval and time interval.  It can't be displayed on a map, but can be displayed by QGIS.