Cluster Analysis

The term Cluster Analysis (Tryon, 1939), which first appeared in statistic as part of exploratory data analysis, encompasses a wide variety of classification algorithms applied in many areas where a general question facing researchers is how to:

• Organize observed data into meaningful structures, which means also data reduction. The idea of data reduction is to use cluster patterns (prototypes) as representatives of cluster members so that an original data set is reduced to a set of prototypes;

• Analyze the structure of phenomena in question;

• Relate different aspects of the phenomena to each other. Clustering allows also inferring hypotheses about the nature of the data that should be then verified using other data sets. The underlying idea is to predict unknown variables using the pattern of the corresponding cluster.

Clustering (sometimes also called numerical taxonomy or typological analysis) is a common technique for data analysis used in many fields such as decision making, data mining, pattern recognition, machine learning, etc. In general, whenever one needs to classify information into manageable meaningful units, cluster analysis is of great utility.

Select one of the topics below to learn more about Clustering with FinMath Toolbox:

• Base Terms and Concepts

• Distance Metrics

• Basics Of Implementations

• Agglomerative Clustering

• K-Means Clustering

• Quality Threshold (QT) Clustering