A Taxonomic Classification Approach for Global Spatio-temporal Data

The World Bank, World Health Organization, and other major vendors provide thousands of country level, time series data sets concerning the environment, health, economics, violence, education, and national security. Practitioners consuming these time series data are informed by and interested in the detection of trends and patterns, keen on knowing how those trends spatially arrange. Trends that indicate economic rebounds, flattened pandemic curves, or steady increases in violence are just a few examples. Given the high dimensionality and noisy nature of the data, it is often difficult to extract and reason about semantic patterns. We propose Categorize Trends, an exploratory analysis tool which labels high dimensional time series data into a manageable set of key behavioral classes and provides the functionality to examine the results spatially using behavioral maps. We apply our method to an important use case exploring the interaction between food scarcity and fertility.