Data visualization can be used to detect hidden structures and patterns in data sets that are found in data mining applications. However, although efficient data visualization algorithms to handle data sets with asymmetric proximities have been proposed, we develop an improved algorithm in this dissertation. In the first part of the proposal, we develop a modified Sammon mapping approach that uses the upper triangular part and the lower triangular part of an asymmetric distance matrix simultaneously. Our proposed approach is applied to two asymmetric data sets: an American college selection data set, and a Canadian college selection data set which contains rank information. When compared to other approaches that are used in practice, our modified approach generates visual maps that have smaller distance errors and provide more reasonable representations of the data sets. In data visualization, self-organizing maps (SOM) have been used to cluster points. In the second part of the proposal, we assess the performance of several software implementations of SOM-based methods. Viscovery SOMine is found to be helpful in determining the number of clusters and recovering the cluster structure of data sets. A genocide and politicide data set is analyzed using Viscovery SOMine, followed by another analysis on the public and private college data sets with the goal to find out schools with best values.
展开▼
