Currently using Map Info Pro V16. I am creating crime hot spot density maps? using the raster hot spot tool. I have not had any formal training. I can create the hot spot map and it looks fine. I would like more detailed info on each step. What is happening in the background. Is there a video that goes into more detail than just select this option. Why do you pick this option etc.
You might find this post by @Dr Neena Priyanka? useful: https://li360.pitneybowes.com/LI360/s/feed/0D58000003oS1BxCAK
There's some additional related information available from these Knowledge Articles too:
A “Hot Spot Density” analysis will create a raster where each cell records the density of samples (or the frequency of events) within a search area.
From a crime analysis viewpoint, you could have a dataset which records the location of crime events in a city over a period of a year. At some addresses there may have been more than one crime event, and this information might also be recorded in this dataset. Hot spot density will give you a visual representation of the density of crime events over the city.
The “Kernel” setting is the most critical. If set to “Data Count” then it sums the number of samples in the search area and records this data in the raster. On any other setting, a kernel density estimator is used. This has the effect of smoothing or shaping the result and the raster may be more visually pleasing. However, only the “Data Count” setting returns data that is analytically meaningful.
If you use “Data Count”, then the “Normalize by area” check box can be set. This turns the count into an actual density in units / square metre (the actual units depend on the coordinate system of the raster).
If you do not select a column then each sample represents one event. If you do select a column, then the sample only represents an event if the column contains valid data. Furthermore, you can set the “Use input column as frequency” check box to weight the sample by the value in the column - which might represent the number of crimes at that location over the time period. It could also represent the dollar value of the crime or the number of people injured in the crime, in which case the raster would represent those statistics.
The “Group by” option is designed to filter the input samples prior to analysis. You could group by the type of crime (which would be recorded with some ID in a table field). The analysis would then return a raster for each type of crime represented in the dataset. Another common operation is to group by time - if it is recorded in the dataset. For example you could generate a raster for each month.
In all cases, you need to define the search area. For each output raster cell, the analysis will search for all samples that lie within the search area centred on that cell. The search area is usually a sphere of a defined radius and the radius can be expressed as a number of raster cells or as a distance. You can also define an elliptical search area with a directional bias, although I struggle to see any application of that to crime analysis.