About Self-Organizing Maps

You can use scatter plots in Results Analytics to view the relationship between two parameters. However, scatter plots are limited to displaying information about only pairs of parameters. In contrast, self-organizing maps reduce high-dimensional data to two dimensions and allow you to identify interesting clusters in a data set and, thus, to identify complex relationships among multiple parameters. Self-organizing maps also help you interpret trends and relationships when you have a large number of data points.

Results Analytics builds a self-organizing map using the parameter values (the values in each column of the data table) for each data point (each row of the data table). A self-organizing map simultaneously looks at all the parameters of all the data points to determine the similarity between the values of the parameters. Results Analytics uses the similarity to cluster data points into hexagonal bins that are arranged in two dimensions. The absolute positioning of a bin based on its X- and Y-coordinates is not important; the information of value in a self-organizing map is found in the positioning of a bin relative to the other bins. This positioning indicates the similarity or the dissimilarity of the bins and, therefore, the points contained in the bins.

By default, Results Analytics displays self-organizing maps for each parameter that you have selected as an objective. If you have not selected objectives, Results Analytics displays self-organizing maps for the first three parameters in the data set. You can display self-organizing maps for additional parameters by clicking Map View Parameters and by selecting the parameters for which to display a map.