Insights through Information Visualization
Inference for QbD is designed to enhance the interactive, explorative and analytical components of process understanding. It is through process understanding that crucial insights are inferred from trends and patterns hidden within product and process information. Such insights must then be effectively communicated to bring about product and process improvements. Information visualization technologies, techniques that support and amplify the communication of data using visual media, are vital to process understanding and improvement.
Inference dfExplorer produces a visual display of 3 data variables as 3-D scatterplots: a collection of points each having coordinates on the x, y and z axes in three-dimensional space. Scatterplots represent the association, not necessarily causation, between three variables. Scatterplots used to explore data, as in dfExplorer, function quite differently than scatterplots used to report on data. Accordingly, Inference dfExplorer has advanced features like those found in Spotfire and Tableau to enable effective and advantageous data exploration. Features include the following:
- Interactive plotting. Depth perception in 3-D scatterplots is dramatically improved by enabling rotation or motion of the point set.
- Point shape, size and outline configuration. Depth cues are further enhanced by these features; especially useful is the ability to vary point size based on the point's distance to the viewpoint.
- 4th dimension display. An additional dimension can be displayed by using assigned colors to points, which is especially useful for a categorical variable.
- Details-on-demand window. Selecting any data point within the scatterplot will display all the attribute values associated with the point in the attributes window.
- Direct, instantaneous data filtering. Adjustment of user-controlled sliders (continuous), check boxes (discrete) and radial buttons (Boolean) enable dynamic filtered of displayed data.
- Simultaneous data views from multiple complementary perspectives. Most insights gained from information visualization come from comparisons. Inference dfExplorer enables comparisons to be made between different values, sets of values, or views of the data by placing several data displays on a single screen.
- Global filtering on multiple complementary perspectives. This feature builds upon the previous feature. Global filters provide the means to remove unwanted data from the entire display. This enables visualization of the effect of multiple variables on each complementary perspective simultaneously.
- Broad data access. Users have access to data from a range of QbD data sources including Inference Excel files, StatDataML files, and data frame objects.
Despite being tightly integrated with each of the Inference Desktop Applications, Inference dfExplorer can be used as a powerful standalone application. Whether used in conjunction with Inference Desktop Applications or as a standalone application, Inference dfExplorer provides capabilities to add, remove, and change views of data on the fly that, when coupled with the power of visual media communication, make it an indispensable partner in the pursuit of process understanding.