Blog 5: Course Summary

Chefs can share their breadth and innovative culinary skills through Menu Dégustation, or tasting menus.  The intent behind these tasting menus is the careful, deliberate delivery of highest quality ingredients to patrons, with recipes ranging from the simple to the complex, in small portions.  Looking back at our Business Intelligence course, each module has been a carefully crafted portion intended to share the variety of concepts and techniques in this field.

Serendipitously, the progression of the course's "menu" followed almost parallel with the needs of a major initiative that I am co-leading in my organization.  Given that this blog is a retrospective of my learning through the course, I will also be discussing my use of each "plate" in the initiative.


Plate 1: Data Warehouse Design

Starting with data warehouse design was a great transition into Business Intelligence, as I had just completed Enterprise Data Management the semester prior.  After a semester of grappling with normalization levels, creating a star schema that on the surface "undo's" all the work of an underlying database really drove home how different architectures are needed for different capabilities.  I was excited to see that Data Warehouse Toolkit was the required reading for this module, as I had recently purchased a copy to read over the summer.  My favorite quote from that book is "QUOTE FROM BOOK", as it speaks to some of the fundamental logic at play when making decisions around data warehouse structures.  The different patterns (inventory, slowly changing records, etc) were all immensely informative as a reference.  That book has since been well tagged, highlighted, and sits on my desk at work.  

My deepened understanding of data warehouse design came at a critical time of my initiative, wherein I was asked to assess the current state data warehouse architecture of the organization and provide feedback on a proposed future state design.  In my role, I partner with the IT department as an owner's representative of sorts, so this type of ask outside my direct wheelhouse is not out of the norm.  Concepts and book in hand, I was able to help translate current state IT to future state business capabilities for our senior management.  While this was only the first step of a longer initiative, I was able to contribute towards successfully progressing this initiative from concept to planning stage.

Plate 2: Data Visualization

I have always loved the duality of technical and creative capabilities at play when looking at a well-done data visualization.  Data visualization has truly become its own art form, with high art being defined as the ability to communicate insights and share information with the viewer.  Before this course, I had only developed in Tableau through an open workshop hosted by another department.  However, by the time this course covered Data Visualizations, my initiative had included the launch of Tableau Server for my organization.  My role then expanded to include the service management of Tableau Server, which meant the organization's report developers would be relying on me to steward a cohesive delivery of data visualizations to consumers.  In learning more about Tableau, and developing dashboards based on the Bird Strikes dataset, I was able to understand the developer's efforts and how to articulate practical design standards for the initiative.  Developing the design standards for visualizations was largely based on the reading Common Pitfalls in Dashboard Design, where I took each common mistake and reworded it into an affirmative aspiration.

Common Pitfall	Design Standard Translation
Exceeding Boundaries of a Single Screen	·       Stay within a Single Screen ·       Test Download as PDF/Image for clean breaks
Supplying Inadequate Context for the Data	·       Assume an uninformed consumer ·       Include reference to access the detailed methodology
Displaying Excessive Detail or Precision	·       Where detailed information is needed, provide a table
Expressing Measures Indirectly	·       Directly express measures
Choosing Inappropriate Media of Display	·       Display media to be additive to the expression of the information
Introducing Meaningless Variety	·       Keep overall dashboard as simple as practical
Using Poorly Designed Display Media	·       Perform quality assurance on visualizations before peer review
Encoding Quantitative Data Inaccurately	·       Perform quality assurance on quantitative data before peer review
Arranging Data Poorly	·       Design for intuitive flow of information
Ineffectively Highlighting What's Important	·       Importance should be highlighted
Cluttering the Screen with Useless Decoration	·       Use Org/Unit brand standard logo only
Misusing or Overusing Color	·       Use Accessible Color Palette and Org color standard templates
Designing an Unappealing Visual Display	·       Perform quality assurance on the overall dashboard before peer review

Plate 3: Website Analytics

Before this course I had only heard of website analytics, but had never used any.  After learning Google Analytics, and applying it to Google's marketplace website for the assignment, I was impressed with the power of this tool.  One of my classmates had a very insightful blog on this module, noting the tension of "the cost of free" with tools like Google Analytics being able to understand so much about website traffic.  Though I used Google's marketplace website for the assignment, I did learn that my organization uses Google Analytics but that the position that performs analysis is currently vacant.  I asked the manager what types of information was developed with the position was filled, and it primarily came down to sharing click-counts by page.  I showed him the possible insights through my assignment, and there is interest in understanding how to use website analytics to better leverage the website.

For my initiative, as mentioned for Plate 2, the release of Tableau Server was coupled with posting certain visualizations on the public website.  While Tableau Server provides basic out-of-the-box statistics on web views of visualizations, the team has decided to define website goals around getting consumers to the analytics page from the landing page.  Using Google Analytics, we will be able to monitor how effective our website design was in getting new consumers through to the visualizations page.


Plate 4: Network Analysis

Ending the course with Network Analysis was fitting, as it was highlighted the value of Business Intelligence - communicating the relationship between data and what can be learned from it.  Prior to this course, I had only used network analysis in a simplistic manner, using the free tool Kumu.  I leveraged Kumu primarily as a visualization tool, to communicate with others the obvious patterns.  Learning Gephi through the tutorials and the assignment, I was able to learn how to derive network metrics - and why.  Though my imitative had no direct component using network analysis, I have begun to pull together a dataset on our stakeholder engagements to prepare an analysis of the effectiveness of our design.  While it is still underway, I hope to be able to use network analysis to show the diameter of the organizational network, to see how many "steps" there are between stakeholders (nodes) of those that were engaged and those that were not.  Hopefully our design will have good reach, and we will see that there are not many steps between these nodes and periphery of the organization.

Conclusion

This course had a lot to share within 8 weeks, and the testing menu approach effectively exposed me to the concepts while encouraging my own continued learning of the tools and techniques of Business Intelligence.