Visualization of Multivariate Health Data using Self-Organizing Maps

Abstract

Data that are multivariate in nature is a type of data that may contain subtle patterns. However, it is considered to be an obstacle in research most of the time since classical statistics may nd it encumbering to analyze. However, computational statistics, a collab- oration between computer science and statistics, o ers a suite of algorithms that may be used to surpass obstacles such as this. The Self-Organizing Map and Data Visualization are examples of these. The Self-Organizing Map is an arti cial neural network that employs a process to reduce multidimensional data into a low-dimensional representation while Data Visualization is a process that aims to give the human brain a visual representation of knowledge about certain data. SOM Visualize is a software that makes use of both pro- cesses. The tool enables users to input data and visualize several patterns such as clusters, associations, as well as a geographical representation that exist in the data. It may give several hypotheses that may be con rmed through other statistical tests and SOM Visualize has therefore enabled the possibility of analysis of multivariate data.