DataExplorerHD v0.1 Documentation

Installation & Run: User Interface:

The overall interface consists of two views and one data operation panel. These visual components are coordinated to provide a comprehensive view of the data by highlighting its various aspects. They are interconnected such that selections and changes made in one component will be reflected in others. The system is designed to be modular and is easily extendable to include additional visual components.

User Interface

Embedding View (a): This is the main canvas of the interface where the results of DR, points embedded in 2D, are visualized. It contains a rich set of user interactions for data exploration. One could apply different colormaps to visualize points by values of a particular dimension, clustering labels or point-wise distortion measures. %In the case of visualizing local distortion, the range of a colormap could be further adjusted by a percentage value to accommodate new values that could be potentially out of range.

Parallel Coordinate View (b): This view displays the original data with each of its dimensions as a vertical axis and each point as a line drawing through each of the axis. A normalization of the range for each axis is optional to increase readability of the data.

Data Operation Panel (c): This panel contains various data operations such as DR and clustering. The panel is part of the interlinked system so that changes made to the dataset are instantly reflected through other views. The panel consists of three sub-panels: The meta-information panel gives a direct view of the data, in terms of its dimensions and statistics, and includes the ability to filter (hide) certain dimensions for analysis; The clustering panel allows the user to select distance metrics, data standardization schemes (see supplemental material) and hierarchical clustering methods (e.g. classical single-, average-linkage), while also allowing loading of existing clustering; and the DR panel enables the user to choose DR techniques and specify their parameters in an online fashion.

File Format:
Data Operations:


Analysis Workflow:

Additional Example: