Exploiting Inter-Sample Information and Exploring Visualization in Data Mining: from Bioinformatics to Anthropology and Aesthetics Disciplines

In this data-overabundant world, revealing and representing comprehensible relationships behind complicated datasets have become important challenges in data mining. This chapter presents recent achievements in applying data mining techniques to two application areas— microarray analysis and anthropology study on Wi-Fi networks—and applies visualization techniques to help integrate heterogeneous databases to obtain useful data interpretation. As the amount of available microarray data has increased exponentially, integration of heterogeneous databases has become necessary. However, direct integration of microarrays is ineffective after normalization because of the diverse types of specific variations in experiments. This chapter reviews two approaches to overcome this issue, and introduces a cube model that combines and outperforms the two approaches by extracting information from yeast genes. This chapter will continue on a recent anthropology study which applies data mining to visualize urban Wi-Fi networks. In the past, artists could not store and handle with huge data without database, and therefore their works typically failed to communicate with other disciplines. Nonetheless, anthropologists have explored implicit relations and viewed them in multiple ways, and they have created a number of different category principles including binary opposition, functional structure and interpretation. These multiple principles can serve as the basis to visualize human thinking and reasoning in cross-cultural and interdisciplinary study. The anthropology study to be described in this chapter will relate Wi-Fi statics in complicated fieldwork databases with easily-understood cultural phenomena. Clear visualizations involve data aesthetics as well, which focuses on how to represent data in eye-conscious and categorized forms. This chapter will manifest that aesthetics can help expand existing data mining ideas to visual representations, with the example of combining the cube model introduced for bioinformatics data mining and the representation of regional Wi-Fi networks in spatial-temporal color charts.