Similarity analysis is classification algorithm used to categorize time series by type. The method utilizes a distance measure known as Dynamic Time Warping (DTW) to quantify differences between time series by shape but not time. In this way, the DTW distance metric captures difference in pattern between time series independent of duration of the form. The DTW distance between members of a set of time series can be organized in s matrix, similar to a correlation matrix, with smaller distance values indicating a greater similarity of form between two time series. Cluster analysis using this DTW distance produces groups of time series with a similar shape compared to other clusters. Both supervised and unsupervised learning applications for classifying time series are supported by this method. In this paper, similarity analysis is applied to supervised and unsupervised cases: (1) unsupervised supernova classification based on how the light produces changes over time, (2) economic analysis of recessions and the following recovery period to identify distinct groups of similar recessions using unsupervised learning, and (3) supervised epidemic classification of the 2014 Ebola outbreak in Sierra Leone to test whether the early data was most similar which of two earlier outbreaks in Guinea and Liberia.