Since the late 1980s, acronyms have inundated the data reporting and data analysis industry during its attempts to define and redefine itself. EIS, DSS, OLAP, ROLAP, MOLAP, KM, KD, BI and others have at one time or another come to the forefront. Granted, they have their differences (to say the least), but what has remained constant is their goal to provide business knowledge.

Figure 1 ranks some common business intelligence (BI) techniques deployed today [reporting, ad hoc, online analytic processing (OLAP), and data mining] by their ease of use and relative business value. Note that as the information becomes more insightful and the relative business value increases, the techniques become more difficult to use. What if the value of data mining and OLAP could be provided in an easy, intuitive, and user-friendly way?

Addressing this conundrum is where PolyVista Analytical Client (or PAC, for this column) clearly demonstrates its power and value. PAC effortlessly integrates the functionality of OLAP directly with the automation of data mining and interactive 3D graphics, as Figure 1 depicts.

Choose a Cube

Starting a new project in PAC begins with the "Select Cube" dialog. Through a database explorer tree, you can choose a list of available Microsoft Analysis Services cubes (the OLAP database engine packaged with Microsoft SQL Server 2000) or a local cube file (*.cub) to use in your project. It shows all registered sources of data, which may include cubes in local or network databases.

In the same way that Excel refers to workbooks and worksheets, PAC refers to projects and workmaps. A project organizes and manages one or more related threads of distinct analyses, called workmaps. A workmap comprises four panes that support analytic functions you can perform on an Analysis Services cube (see Figure 2):

• Dimension Tree: selectable and manipulatable measures and dimensions
• Visualization: selectable 2D and 3D data visualization
• Table (grid): 2D grid to create and display tabular data
• Analysis: selectable data mining algorithms.

Discovering the Value

The Data Pane is in the upper left quadrant of the application window in Figure 2. It displays the Dimension Tree, which serves as the query engine against the multidimensional cube you're analyzing. You can view and navigate dimensions in either the default, parent/child, or hierarchical mode, or in the dimensional-level view.

This pane includes two wizards: one to create filters and another to create calculations. You can apply filters to individual dimension members, excluding their values from the resulting data you analyze and display. Calculations can create new calculated measures or dimension members. You can drag elements that you want to present directly from the Dimension Tree to the Visualization Pane or Table Pane.

The Visualization Pane is in the upper right quadrant of the figure. Visualizations include a mixture of more than 30 standard Excel-like charts and graphs as well as fully interactive 3D graphic displays. The various visualizations interact with the Dimension Tree supporting the drag-and-drop functionality. They also synchronize with the table so that highlighting a cell in the table will also highlight that cell value in the visualization, and vice versa.

A visualization type worth some consideration is the Pixel Map. When complex queries result in many rows and columns, interpreting those results can be problematic. When the table is too large to see all the cells at once, the Pixel Map renders an image of the table where 100 percent of it is visible in the Visualization Pane. By default, the Pixel Map highlights the cells that represent the top 5 percent in green and the bottom 5 percent in red. Other values are highlighted in gray, and nonexistent values remain white. Moving the cursor within the Pixel Map causes the table to auto scroll to keep the displays synchronized.