The Four Generations of Analytics Tools

As the science of analytics advances, new tools are being developed to allow analysts greater abilities to access data and unearth information. The “state of the art” continues to be a moving target, and analytics tools continue to advance as the skills of the analysts increase. The more recent tools require a broader skillset that incorporates both “domain” knowledge about the subject area and technical knowledge of how to use the tools. In well-trained hands, third and fourth-generation analytics tools can empower analysts to perform projects that would previously have required a team of programmers and domain experts using conventional database SQL-like queries, exporting data into spreadsheets for final analysis. Now users can utilize the same tools to extract and enhance data, enhance models themselves, and perform “stream of consciousness” analyses that would have been impossible only several years ago.

First Generation tools include “static” reports, or reports that allow only filtering of one or two characteristics of data. These reports may be produced in PDF files, through a reporting system such of SQL Server Reporting Services, or through conventional Excel models. They frequently use dropdown boxes to allow filtering a large set of data by some characteristic, such as a hospital name or geographic area. Results are presented in a tabular table, such as the one shown below. In some systems, no additional data manipulations are possible, and data must be exported into Excel for any further analysis. An example of a first generation model is below, showing the selection list that would be used to filter the table. These types of tools are characterized by the fact that the columns to be placed into the table, and the data model from which those columns are extracted, cannot be modified by the user.

Second Generation tools include “dashboards” that still use tabular tables to report data, but allow additional flexibility to filter data in the form of buttons or “slicers”. Often internet-based, they return results to a web browser, which would require exporting and downloading for further analysis. In some cases there is interaction between different parts of a dashboard; for example clicking on one bar of a bar chart might filter a nearby line graph. Dashboards are useful for users having limited analytics skills and who want to monitor a specific set of metrics over time. An example of a Second Generation dashboard for bundled payment analysis is shown below, displaying the costs by data type and timeframe for the DRG selected in the bar graph at the left of the screen.  While different DRGs and time periods can be selected as “filters” for the data, the data elements presented on the dashboard, and the data model from which those columns are extracted, cannot be modified by the user.

Third Generation tools generally utilize a pivot table interface that allow users to build an analysis from a list of available of available data elements. These models have no inherent structure (i.e., they aren’t limited to specific rows, columns and filters) and allow (in fact, they require) the user to specify the data elements to appear at various locations of the table. This requires a more skilled user who is familiar with all of the data elements in the model and how they interact, but it also provides that user with significantly greater power to shape the analysis as desired. In the example below the user has selected the DRG and DataType fields for the row and column headings, and has chosen to summarize the Average Payment amount by those fields. She has also filtered the Episode Families and by the year 21013. The underlying data model cannot be modified by the user, however.

Fourth Generation tools are an extension of Third Generation models but also allow the user to modify the underlying data model. This allows extending the data model by adding new data sources, creating ad-hoc hierarchies of data, creating relationships between data sets, and adding calculated measures and data elements from the original data elements in the model. This gives a knowledgeable user significant power to quickly design analyses whose only limits are dictated by the available data.

 Singletrack Analytics and our partners utilize Microsoft’s PowerPivot models to provide Fourth Generation flexibility for our clients.