A new term, a new course. We kicked off the Applications in GIS course learning about how GIS can be used in crime analysis. We made a choropleth map and a kernel density map depicting 2018 burglary rates in Washington DC. Then we created 3 hotspot maps (Grid Overlay, Kernel Density and Local Moran’s I) using 2017 homicide data for Chicago. We determined which analysis method was a better predictor of areas at higher risk of future homicide when compared to 2018 Chicago homicide data.
Crime density was highest within the Grid Overlay map, and
lowest within the Local Moran’s I map. There was a correlation between crime
density and total area of the of the hotspot, with Grid Overlay having the
lowest total area since we set our parameters to consider only the top 20% of
crime areas.
I did make an obvious mistake when I initially created my
Grid Overlay map. I calculated the top 20% manually since there isn’t a tool
for this task. There was no way to filter by attribute for this task either, but
initially I used the select by attributes option and selected by objectID the last
62 features. I realized when I was completing my write up that even though I
had sorted by homicide count in my attribute table, the object ID was a
standard and would not have changed when I chose to sort by homicide count.
There isn’t an option to select by the field number on the far left which would
have been an accurate choice. I suppose I could have added a new field and
numbered them that way but it was quicker to just manually select from 249 on
towards the end after they were already sorted by homicide count. This was an
obvious error, and I wasn’t sure how I overlooked it but fortunately I caught
it because it would have greatly skewed my results. My initial (wrong) Grid Overlay
had a significantly lower density than the other two analysis methods. One fact
to consider when running these analyses is that the more manual tasks are
involved in an analysis method, the more room there is for error.
Here are my 3 hotspot maps for the city of Chicago.
No comments:
Post a Comment