Chapter 4: Identifying Clusters
Chapter 4 explains how to identify clusters, which happen when features are found in close proximity. By pinpointing, it can help to determine cause of clusters in that particular location. Statistics are used to determine if there are reasons for the clusters or if they happened by chance. Cluster of features with similar attribute values can be brought up using discrete features, spatially continuous data or data summarizing. They are interval or ratio values. Depending on what you are trying to pinpoint, you may have to put a specific period of time, or even a specific date. Clusters are almost always defined by straight-line distance. This could work, unless you are trying to find distance in travel time. The nearest Neighborhood hierarchical clustering specifies the distant features that can be found from each other in order to pbe part of a cluster. It also determined the minimum number to be able to consider it a cluster. It can also show the clustering at different geographical scales. To see the orientation of individual clusters, GIS may calculate the standard deviational ellipse for the points. To find the causes of clusters, you would want to compare clusters to a control group. To do this the control group and clusters can be mapped together, or you can creat clusters for the control group and compare them with the original clusters being analyzed. Clusters can also be identified on whether they are similar to their neighbors or not. Basically, if high values are surrounded by high values, they were similar, and vice versa. Using Moran’s I, means you are interested in local variation. A large positive value for Moran’s I indicates that the feature is surrounded by features with similar values, and a negative value means the feature is around dissimilar features. The G-statistic shows where cluster of high and low values are. There are two different methods. The Gi statistic helps you determine the effect of the target feature and what is going on around it. Gi* is the where you can find hot and cold spots.
Chapter 5: Analyzing Geographic Relationships
Chapter 5 begins with examples of how GIS is used in different fields. Some of the fields mentioned were transportation analysts, environmental lawyers, archeologists, state police, and wildlife biologists. In stats, attributes are the variables. Two analyze the relationships between attributes, you can use a defined area, sample point, or raster cell. The variables from different layers need to be associated with the same geographical unit. A ratio needs to be used if the locations are two different sizes. For different sets of features, they need to be combined somehow. To do this, you can either do a polygon layover or create rasters of the areas, making them the same size. Variables can also be created to represent spatial interaction between features. These features could be distance, travel time, or travel cost. Statistics is a huge thing when coming to analyzing geographic relationships. spatial autocorrelation is one. It violates the assumption that observations are independent. It brings the redundancy into analysis. To study the relationship between two variables as well as the nature, you measure the extent to which they vary together. Values have a direct relationship is the they both increase when one of the also increases. If one decreases while another increases, that is an inverse relationship. Other than that, there won’t be a relationship. There are also posiitive and negative correlations.