Pattern Analysis

Assessing the distribution patterns of Emergency Calls in Battalion 2 and Patron Locations from the Oleander Library.

Problem Statement

The Fort Worth Fire Department receive a lot of random emergency calls in the Battalion 2. Nevertheless, it has limited understanding of whether the calls are clustered or dispersed or are randomly located. It has data which shows different incident types and the priority rating value of the incidence. It aims at stationing emergency intensive care units near hot spots, hence asked for help in spatial pattern analysis. The objective of this assignment is therefore to assess distribution patterns of Emergency Calls in Battalion 2 and Patron Locations from the Oleander Library

Analysis procedure

To address the problem, I used ArcGIS Pro 2.9.4. The problem was solved using three main strategies, 1) I used Average Nearest Neighbor tool to analyze the data and determine the degree of clustering. Data used was the EMS Calls –Feb 15 data, Battalion 2 Boundary shapefile, and the Incident type layer for 15 February which were taken from the tutorial 8-1 map package. Required readings were pages 63-79 and 88-96 in The Esri Guide to GIS Analysis, volume 2. This helped with improving my understanding of the tools used to solve the problem. 2) I used the High/Low Clustering (Getis-Ord General G) tool to evaluate if there was clustering of the values as well. Data used was Calls For Service for Jan 15 layer, Battalion 2 boundary, all which were taken from the Tutorial 8-2 E map package. In preparation, I read pages 104-108 in the Esri Guide to GIS Analysis, volume 2. 3) Then I used the Multi-Distance Spatial Cluster Analysis (Ripley's K function) tool to determine whether clustering will still occur when a single feature is assessed against relative distance of multiple features not just its neighbors. Data used was calls for service data for January 15 layer, the Battalion 2 boundary layer, all which were taken from the Tutorial 8-3E map document. 4) Finally, I further to assessed whether the clustering can be observed using a combination of both feature locations, and feature values. For this, I used Spatial Autocorrelation (Global Moran’s I) tool. Data used was calls for service February layer, 500 ft Grid layer, from Tutorial 8-4 map package, and also used Patron Location Layer, 300ft Grid layer, from Tutorial 8-4E map package. In preparation, I read pages 80-87 and 118-26 in the Esri Guide to GIS Analysis, volume 2.

In part one of the assignment, I used Average Nearest Neighbor tool to determine if the Emergency Calls for Service were just random or had a clustering pattern. Since the tool I used is sensitive to the area of study, I used the Battalion 2 layer to restrict the tool to its square feet area. After running the tool, it had a Z-score of -18.810024, with a significance level of 0.0. The nearest neighbor index was 0.549. With this, I rejected the null hypothesis and concluded that the data did not show any pattern of randomness, but was clustered. For the exercise, I used incident type data, with a definition query greater than or equal to 700, but less than or equal to 745. This produced a Z-score of -1.7 and an index of 0.914. I rejected the null hypothesis and concluded that the pattern of the incidences was not random.

Second part was to look into the values themselves, to see if the values of the call priority were just random or not. I used a High/Low Clustering (Getis-Ord General G) tool on this. But the tool required particular distances to be tried out in the assessment. I used the Calculate Distance Band from Neighbor Count Tool to determine average minimum and maximum distance at which each feature can find 7 neighbors around. Average distance was found to be 1001. Then, I used a range between 1000-1400 to calculate the Getis-Ord G index. But for the Calls for Service in January 15 layer, I used the distance range of 700-1200 feet, and maximum z-score was found at 900ft, with a z-score or 9.401. I rejected the null hypothesis and concluded that the calls formed a cluster pattern.

Third part involved looking at both values and location of features to determine if the calls were just random or not. I used Multi-Distance Spatial Cluster Analysis tool. I used Calls for Service in January layer as an input layer, with a minimum distance of 200, at 100 intervals, with 10 bands. Then created a new field of difference in the K function January 15 table, which calculated the difference between the High Confidence Envelop and the Observed K value. This field was then plotted on a graph to identify the distance at which the clustering is more pronounced. This was after the Ripley’s K index which was high enough to help with rejecting the null hypothesis that the calls were randomly distributed across the Battalion 2.

Lastly, I used spatial join to join the point Patron Location from the Oleander Library layer with a 300ft grid layer. I conducted a definition query to remain with grids which had a joint count of greater than 0 features. I used Spatial Autocorrelation (Global Moran’s I) tool which uses both values and feature location to determine the data pattern. I tried distances ranges of 2800-3800 ft with 200ft increments. The highest Z-Score was observed on 3400ft with a 12.059 z score. I also rejected the null hypothesis and conclude that the Patron locations were not just random.

Process Diagram

Results

Application and Reflection

Problem Statement: With about over six million cases of malaria and 2,000 deaths reported each year, Malaria is one of the leading causes of morbidity and mortality in children under five. USAID/Malawi would like to help to reduce the prevalence and child mortality, but firstly has to establish whether the cases are just random of form a spatial pattern for targeted interventions.

Data needed: Malaria incidence geocoded data. The data will be obtained from USAID/Malawi GIS database.

Analysis procedure: I will use the Spatial Autocorrelation (Global Moran’s I) tool to assess whether the Malaria incidence cases in Malawi are random or form any clustering patterns. I will first assess the average square kilometers in Malawi villages (Administration boundary level 4) to determine the grid size to use. Using spatial join, I will join the grid data to malaria incidence health facility point dataset. Then will run a definition query to make sure I rule out all the grids with no data. Finally will run the Spatial Autocorrelation to determine the spatial pattern of the data.