Final Thoughts

                The past eight weeks have surprisingly flown by. I came into this class expecting something exceptionally difficult (based on MIS 531 being a prerequisite) and I was pleasantly surprised to instead find a thoughtful, engaging class waiting for me. Although not entirely what I expected MIS 587 has been an interesting ride and I’ve learned quite a bit, both new information and solidifying concepts I was already aware of from work and other classes. In the paragraphs below, I’ll outline some of what I have learned, how I think I can applying in my career, and general thoughts about the class and information imparted. I will say for the record that this portion of the class, the social media engagement, has been a unique experience. We have an expectation as students that there will be some discussion component to any course (and especially in an online course, a written discussion component). Using the standard D2L message board is boring. Forcing us all to engage on social media however is interesting; It gets us out of our comfort zones. We all exist on social media (I’m sure there’s an exception or two to make the rule) but much as we compartmentalize our work selves and home self’s, we also compartmentalize our school selves. Exposing our social media habits to each other was a fun exercise in driving real engagement. Sadly, I’m not sure it had the intended effect: we all engaged, but the minimum amount. Few people responded to the comments left on their twitters and fewer still to comments left on each other’s blogs. If the goal was to foster better, more open communication I think we fell short of the goal. Also, having us share articles and blogs on social media was a missed opportunity to have a final Assignment really relevant to the material: using Gelphi to analyze our own network of interactions. I thought for sure this was the point of the whole social media exercise and was shocked when the final Assignment was different. Perhaps we simply didn’t interact enough to make it worthwhile, but even still using our own connections to power a Gelphi analysis of networks would have been fascinating and very prescient. Perhaps in the future that will be the final assignment.

                Now going module by module, we start with the Dimensional database and Star Schema. Coming directly from MIS 531, where we literally built a Relational Database from scratch pivoting to a different, potentially better way of making a database was a nice touch. I had recently learned about NoSQL databases at my job (at the time) so I was interested to see them in action and it did not disappoint. Relational databases made sense; reduce the housed information as much as possible, create new tables as the data is normalized to reduce issues. But the deeper you go the bigger the database gets, and query time begins to become problematic. Case in point my previous job, with massive queries used to hit a 6 terabyte database and always plagued by performance issues; there had to be a better way. Star Schema provides that better way, in certain circumstances. Although not a blanket replacement for relational databases, depending on the type of information needed to be stored the reference fact table approach can greatly reduce the time to pull up information. The design of the database is also much, much simpler. Sadly, my previous employer’s information is much too large and diverse (multiple school districts storing hundreds of columns for students) for this to work but my new job (I switched companies in the middle of the class) has potential. Just the other day I was musing about how using a dimensional database could significantly improve system performance in certain client settings. The potential applications are definitely worthwhile for this type of database versus that standard bloat of the relational.

                The next module was my clear favorite, creating Dashboards on Tableau. I’ve always been intrigued by the various dashboard designs our there, starting with my firs professional job at the start-up LivingSocial. I worked as a point of contact/tech support associate for the merchant partners who sold promotions via LivingSocial’s website and email list, and much of my time was devoted to educating merchants about and troubleshooting the Merchant Center, LivingSocial’s portal for merchant partners to manage their promotions. This portal of course contained various metrics regarding the promotions as envisioned through a dashboard and various widgets. I was always curious about the science and structure behind them; how they were made, what the process was, how to implement them. When I moved on to my next job doing QA I got more first-hand experience, testing the widgets being implemented for a new Dashboard project for the company’s clients. But it wasn’t until the module of this class that I got to try my own hand at Dashboard creation. The amount of thought and planning necessary to create a functional dashboard was a surprise; judging by the developers at my previous job I assumed it to be an ad-hoc, tossed together process. When faced with crafting my own dashboard however, it took time, planning and thought about how to convey specific information in the best possible way/ It also required viewing both the data and business or client needs holistically as opposed to just thinking how to display the data. My Dashboard consisted of four widgets that had little overlap but worked in tandem, each providing unique information by the same variables across all widgets. In that way I ensured the viewer would be able to see the data from multiple facets by multiple variables and I was very pleased with my results. In fact, Dashboards have been on my mind so much that in my recently started job I’ve been musing about making changes to my Scrum team’s TFS dashboards to see if we can get some more interesting visualizations. The visualizations the team currently uses are fairly basic, covering burn rate or each sprint, open unassigned user stories, etc. But given the potential uses of TFS I’d like to experiment, putting my Dashboard creating skills learned in this class to use making something I can really use as a Quality Engineer for tracking sprint progress.

The Google Analytics unit was interesting as I have used it extensively in my previous job, and I discussed my usage in my blog post for that week. Because my usage had been very limited to essentially using GA to gain client behavior insight regarding device usage and popular pages, the module on GA for this class was very useful in fleshing out the parts of GA I had never used. The Assignment really challenged me to put my thinking cap on and dig into what GA can tell us about a website’s functionality and if page flow and design is sufficiently intuitive.

Lastly, we used the Gelphi tool to analyze a social network. I had actually used Gelphi before in the MIS Online Data Mining class which I took last summer, although that usage was much more structured and guided versus that “have at it” approach in MIS 587. I actually preferred the more directed, controlled approach from Data Mining, as unless one is specifically a Business Analyst or Data Scientist these concepts and the usage of Gelphi can be complex; most of us are not and are very new to these concepts. I think there’s room for much more guidance in utilizing the program versus just telling us to explore and come up with some form of analysis. Perhaps adding another module exploring not just the nature of network analytics but the actual features of Gelphi would have been very helpful in completing the network analysis assignment. The tool itself is very useful, and as I mentioned in my intro I would have been very excited to analyze our own network data from our social media interactions and see how the network comes out. It would be interesting to see which of use commented the most, how communities formed, who had the most betweenness, and digging deeper, what specific skills/job/education those individuals have versus others. As I mentioned in my first blog post, my mind always reaches back to my social sciences background to consider how big data analytics can be used to further our understanding of humanity, and this is a prime example. Social media influences could potentially have a correlation with certain careers. They could be more outgoing, or not! The excitement comes from digging into the data, cross referencing it with personal metrics, and seeing the holistic picture of the individual pieces of the network. The data paints a picture alone certainly, but when you ground it in context you gain incredible insight. Understanding the network connections is one thing (and it’s useful) but if you then understand the individual nodes on a deeper, contextual level, the data because a powerful facet by which to understand the world.

So in (brief) closing, I learned a heck of a lot in MIS587. Some things were more useful (to me) than others, but all of it was worthwhile and most importantly, applicable. As data becomes more and more ubiquitous and we tap into it the need for analytics, dashboards, and metrics will increase exponentially and the potential here is rather limitless.