UC Santa Cruz
Interactive High Dimensional Data Analysis using the Three Experts
- Author(s): Albrecht, Georg Hans
- Advisor(s): Pang, Alex
- et al.
With the increasing availability of different kinds of data from various domains such as health care, finance, social networks, etc. there is a need to provide analytic tools that are more accessible to lay people. In this paper, we present a software tool which can be used to help make high dimensional data understandable for inexperienced users. To facilitate the understanding of the data, we place special emphasis on how the data is presented, using the ``Three Experts", and on showing personalized information within the data. The Three Experts display shows the results of three different dimension reduction techniques, similar in notion to seeking several expert opinions regarding a particular topic. This will help the user to discern between pertinent structures in the data, and those resulting from the distortion inherent in dimension reduction. The second emphasis is on providing the ability for users to identify, insert, and manipulate points of interest among the sea of data. In addition, the user can observe high dimensional trajectories from one position in the data set to another. This will help convey the changes necessary to displace a point to its desired target state. Observing these changes will enable the user to develop an actionable intuition for the data in question. Though the currently envisioned application for such a system is in health care, such methods could potentially be used in any field where high dimensional data needs to be analyzed.