Online Program Home
My Program

Abstract Details

Activity Number: 259 - SPEED: Environmetrics: Spatio-Temporal and Other Models
Type: Contributed
Date/Time: Monday, July 30, 2018 : 3:05 PM to 3:50 PM
Sponsor: Section on Statistics and the Environment
Abstract #332921
Title: An Application of Monothetic Clustering to Data with Circular Variables
Author(s): Tan V Tran* and John C Priscu and Mark Greenwood and Marie Saback√°
Companies: Montana State University and Montana State University and Montana State University and University of South Bohemia
Keywords: monothetic; clustering; circular; antarctica
Abstract:

Monothetic clustering is a clustering technique that has advantages in interpretability and prediction compared to other clustering techniques. It is based on recursive bipartitions of a data set by choosing splitting rules on the variables, one at a time. Circular variables, the type of variables whose values are defined within a range of values where the upper bound coincides with the lower bound (e.g., directional variables, time of day, etc.), must be treated differently from conventional quantitative variables. We suggest a clustering procedure for a data set that contains circular variables, starting from visualization, to calculating an appropriate distance matrix, to a clock's hand-type partitioning used in monothetic clustering. The suggested method is applied to an air particle counter data set measured in Antarctica. Using data measured every 15 minutes, the particle counts, wind speed, and wind direction are used to create meaningful rules to partition the multivariate data set. By doing that, we can identify groups of conditions for different patterns in sediment transport in the Taylor Valley in Antarctica.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2018 program