Analytics

Overview of clustering/other ML methods used.

To analyze the death rate(s) in a meaningful manner, we decided to cluster our data. For clustering, we ended up using k-means clustering, although diana was also shown to be rather effective for some parts of the data. There is a clustering function for both k-means and diana. K-means basically tries to group data points into clusters that are the shortest distance from the other points in the cluster. Diana clusters by starting out in one giant cluster and then attempting to partition the data. Both results produce similar numbers of clusters. If a state has six or less counties (Delaware, Hawaii, and Rhode Island), then it was not clustered at all and became an exception case. Exception cases were handled separately in parts of the app. For example, the line chart shows the mortality rate by county and not the mean mortality rate for cluster (both are over several years). The factor distribution maps also do not exist because there are no clusters for us to map. The box-plots become a single line (there is only one value per social determinant per county, so this is mathematically correct) and the dotplot below it also does not show clusters (the clusters do not exist).

To analyze our "Social Determinants" in a meaningful manner, we used a Kendall Tau test. A Kendall Tau test is typically used for non-parametric rank correlations. It basically tests whether there is an association between data and returns values from -1 to 1. In this case, 1 means the variable is associated with higher mortality rates and -1 means it is associated with lower mortality rates (or rather the clusters for higher and lower mortality rate respectively). The higher the absolute value, the greater the association. Essentially, we are testing what medical, economic, and social factors are related to our higher mortality rate clusters (or counties, in the case of the exception states).