Module 1 | Data Warehouse Design

Module 1 has covered quite a bit of ground on data analytics in the context of Business Intelligence. Its content covers a few main topics, namely, Data Warehouse Design & Star Schema Design, Performance Management & Balanced Scorecard, Data Quality Analysis, and Dashboard Design and Analysis. This article overviews the four questions.

Data warehouse Design & Star Schema Design

The module started off by introducing the notion of data warehouses and data marts, and the data design system known as Dimensional Modeling (Kimball et al. 2016). The essential protocol for dimensional modeling involves (1) defining the business process, (2) declaring the grain of the model (i.e., events where data are collected from), (3) identifying its dimensions (i.e., its context), and (4) identifying its measurable facts.

An important characteristic of dimensional modeling is its support of database design where database normalization is largely avoided. As Kimball and Ross (2013) put it, “the price paid for greater analytic flexibility is often greater complexity. Although IT professionals may be impressed by elegant flexibility, business users may be just as easily turned off by complexity.” 

I see a tradeoff which seems to be analog to that of code writing and web content development. Basically, two paradigms seem to go opposite directions:

• Minimize information redundancy. This approach aims to protect content integrity, reduce maintenance time, and minimize storage of information, at the cost of increasing the complexity and the number of connections between items (whether they are data objects of modules of source code).
• Allow for redundant information, to make readability of content more user-friendly, improve access of content by certain bits of information (e.g. queries or functions in source code), and thin the network of joins between bits of information. (e.g. this is found in both code development and database modeling).

In general, a model staying halfway of these theoretical systems may be the preferred practice, to facilitate maintenance and update of content but also enable a logical structure for efficient access and navigation of the data space.

 

Data Quality Analysis 

This section was the shortest in the module. Its importance to guarantee a reliable and efficient data warehouse is, however, unquestionable. Most of its content revolves around data profiling and data governance. In data profiling, the goal is to discover and understand the content, structure and quantity of data, as part of managing and governing data. Examples of profiling techniques include visualizing the frequency of values, uncovering associations between data attributes, and accurately depict the completeness and consistency of attribute values. In my search for resources on data quality management, I found the following publications to be a good source for definitions and method:

•  Panoply’s Data Warehouse Guide
•  A Classification of Data Quality Assessment and Improvement Methods (Woodall, Oberhofer & Borek, 2014)

  

Performance management & Balanced Scorecards 

The module also focused on particular methodologies for the management of business performance. Concretely, it considered the Balanced Scorecard (Kaplan and Norton, 1996; Schneiderman, 2006). The Balanced Scored card includes goals and performance measures that respond to four perspectives: financial (e.g. sales, revenue), customer (e.g. customer satisfaction, customer purchasing preferences), internal business process (e.g., overall cycle time, product quality), and learning & growth (e.g., life cycle of a product, design of new products, employee training). The balanced scorecard has evolved into a second generation and a third generation to improve the selection of more relevant measures (Lawrie and Cobbold, 2002). I found it interesting to note how the Balanced Scorecard concept, although popularized by Kaplan and Norton’s publications, dates back to the mid-1980’s and originated from a professional business environment, just to be discovered, modeled, and adopted later in academic management thinking (Schneiderman, 2006). 

I seem to see capacity in the Balanced Scorecard to define the dimensions of the search space in multiobjective optimization problems. Provided, alternative strategies or several cases can be compared (e.g. different branches, years, regions, or simulated business models) a global measure of comparative performance (e.g., a Pareto front) could be added to the framework to assist managers in assessing and prioritizing objectives based on a population of cases.

Dashboard Design & Analysis

The basic notion about a dashboard is that we are using a graphical environment that helps to visually describe key performance indicators at a glance. Few (2006) points out that, despite the existence of suitable graphical user interface technologies in the, dashboards didn't become widespread until the late 1990s thanks to the popularity of KPIs and the adoption of Robert S. Kaplan and David P. Norton's Balanced Scorecard (Kaplan and Norton, 1996). We have introduced the Balanced Scorecard above. The graphical analogy and complementarity between both tools are evident, but conceptual differences can be identified: while scoreboards are more of a tactic tool to monitor performance, a balanced scoreboard is a strategic framework where quality indicators of performance are prescribed.

Multimedia (text, image) and multimodal information formats (chart, map, measures) provide a visual interface with capabilities to interact with dynamic data. Although not exclusive of dashboards nor caused by them, miniaturization of multimedia devises (tablets, mobile, laptops) may affect the communicative capacity of dashboard, as graphical features in the dashboard need to be compresses, simplifies, and minimized to makes to prioritize information at the same time that saturation of information in the screen is avoided.

 

References

Few, S. (2006). Information Dashboard Design. Reilly. [URL]

Kaplan, R.S. and Norton, D.P. (1992). “The Balanced Scorecard – Measures That Drive Performance”. Harvard Business Review. January–February: 71–79.

Kaplan, R.S. and Norton, D.P. (1996). “Using the Balanced Scorecard as a Strategic Management System”. Harvard Business Review. January–February: 75–85.

Kimball, R. and Ross, R. (2013). The Data Warehouse Toolkit: The Definitive Guide to Dimensional Modeling (3rd ed.). Wiley.

Lawrie, G. & Cobbol, I. (2002). Development of the 3^rd Generation Balanced Scorecard: Evolution of the Balanced Scorecard into an effective strategic performance management tool. 2GC Working Paper.

Schneiderman, A.M. (2006). "Analog Devices: 1986-1992: The First Balanced Scorecard". URL: https://www.schneiderman.com/Concepts/The_First_Balanced_Scorecard/BSC_INTRO_AND_CONTENTS.htm [Accessed: 9/20/2020

Woodall, P., Oberhofer, M., and Borek, A. (2014). “A Classification of Data Quality Assessment and Improvement Methods”. International Journal of Information Quality, 3. DOI: https://doi.org/10.1504/IJIQ.2014.068656