M4 - Crime Analysis

-

I am one of the discussion leader for this module as I'm very familiar with the use of kernel density analysis. I had used this to compute for water table of are area of interest based on data points of water table in a series of sampling area.

For this module, we focused on Chicago's Homicide Crime Data for 2017 and then did a statistical comparison for 2018. We used 3 different types of hotspot analysis named, Grid-based Thematic mapping, Kernel density and Local Moran's I. 

Grid-Based Thematic Mapping 

We had spatially combined grid data over crime data and then summarized the numbers per grid cell. We only selected the top 20% of the overall grids with highest point count per grid.

Kernel Density

For this analysis we did not need a bounding area to spatially join our crime rates as it calculates the density of features in an area. This considers the occurrences of crimes that could put additional weight to an area when producing a hotspot for a certain radius or cell size.


Local Moran's I

This is a method I haven't used before as it uses a complex formula that computes hot spots, cold spots and outliers based on Local Moran's I index. Then a cluster/outlier type will be assigned per census tract that consists of high values (HH), low values (LL), outlier of high value surrounded by low values (HL) and outlier of low value surrounded by high values (LH).


Lastly we used the following results to compare the hotspot maps with the locations of homicides in 2018. The following table below was calculated to help us measure how useful our hotspot map is predicting future crime. 

Hotspot Technique

Total Area (mi2) in 2017

No. of 2018 homicides within 2017 hotspot

% of all 2018 homicides within 2017 hotspot

Crime density (2018 homicides within 2017 hotspot per mi2)

Grid Overlay

15 sq mi

159

27%

10.6

Kernel Density

25 sq mi

256

43%

10.2

Local Moran’s I

35 sq mi

270

46%

7.7

In conclusion, I would choose the Kernel Density analysis as the most efficient. One is that the combination of the % of all 2018 homicides within 2017 hotspots and crime density are both high in the said method. Second, the kernel density provides a much more detail point aggregation instead of grid coverage. Lastly, it pinpoints specific hotspot areas where the police chief can focus his resources. The other 2 provides a larger aggregation via grids which can be hard when the police department has limited resources.


Comments

Post a Comment

Popular posts from this blog

M1 S4 - Suitability Analysis / Least Cost Path

-
Image
Scenario 4 lets you pretend to be a park ranger in the Coronado National Forest and is assigned to create a corridor between two protected areas for black bear movement. This will help us keep the area protected on different factors especially urbanization keeping away people and noise away from our black bears yet giving them the best suitable area to live. Below is the ArcGIS model I created for the said scenario. We are given 3 factors to consider (1) roads, (2) elevation and (3) land cover. I realized that besides these if we have data of rivers, and if lucky GPS tracks of black bears (if that's even possible) so we can get a glimpse of their movement as well. First, I applied Euclidean distance tool with the roads layer to transition it to raster and then reclassified all of them according to the suitability criteria given. The reclassified layer were then combined using Weighted overlay tool with the following weights: land cover (60%), elevation (20%) and roads (20%).  Moreo...
Read more

GIS Portfolio

-
For the last few weeks of my certification in UWF, I have been working on my GIS portfolio on WIX website. This GIS portfolio showcases an in-depth about myself and some of my GIS projects that I've worked as a student and as a GIS profession in our organization. I had a hard time digging up some of my works on the past years when I was just starting as a consultant. I realized that I have improved a lot in my skills and this is actually a good avenue to showcase some of it. Here's the link to my GIS portfolio . In addition, I also am preparing myself to apply for the Master's degree program in GIS Administration in UWF and continue my study. I hope that this portfolio will be helpful if needed be as a requirement in my application. I'm also sharing my LinkedIn here that contains some of my works and some of the articles I've written with some of my colleagues:  How to Protect Refugee Camps from Floods and Cyclones Using Geographic Information Systems Digitization ...
Read more

Module 4 - Geoprocessing

-
Image
Geoprocessing is a an importent component in GIS. Whether you're a new or advanced user it is an essential set of tools you will use for processing geographic and related data. It offers to perform simple spatial analysis to complex ones. It can also be modeled into a sequence of spatial relationship that will help the user to automatize tasks from one tool to the other and produces a whole new dataset you can use for decision making. Another component related to geogprocessing is the Model Builder, which I recently have used quite frequently this year. In one of the projects that I'm working on we had to run multiple tools from the input until the end result. And we need to do that everyday, instead of running 10+ tools daily, why not create a model wherein all you need to do is add the input parameters (daily) and run it. It will then publish the end result in my ArcGIS Online profile and there you go I now have an updated dataset I can publish daily. Model Builder is lik...
Read more