Week 4 has definitely opened my eyes to the necessity of r the BI discipline and even the Data Scientist career path. After spending the first few weeks on data-warehousing, it didn't really feel all that different than an extra layer on top of a strong knowledge of SQL and relational databases. Sure, the Star-Schema thing is a little different, and the ability to break the normalization rules is no skin off of my back (and I imagine the backs of others too).
Even after reading Lecture 8 on dash boarding, it seemed as though it would be easy to jump right into a BI tool such as Microstrategy. In fact, I was taken back by the need to take 23 quizzes on using software. The tutorials and quizzes were an easy introduction to the tool, but it wasn't until beginning the second assignment that I began to realize how important and "heavy" business intelligence really is.
Nonetheless, I dove in fumbling through the Microstrategy interface. It seemed pretty easy at first and like any other visual reporting tool (Crystal Reports comes to mind.). However, as the realization hit that I was being tasked with scouring a significant amount of data, identifying key attributes and metrics, and merging it all into something that my boss could read and even understand-- that's when it became apparent that this is no easy task.
The first question was really not that challenging with the exception of building the familiarity with the tool. Working on the second question was where it became more challenging and more indicative that either I wasn't getting Microstrategy all that well, or maybe Microstrategy is an ok tool, just not a great one.
I was able to generate custom reports that had some meaning fairly easily, and I was able to create what Microstrategy deems a dashboard/scorecard easily as well. Although, after reading Tyler's comments I questioned what I had done. After watching Hans Rosling's TED talk, it was clear that dashboarding is as much about data analysis as it is communication in general.
After working pretty intensely on Part 2 of the assignment, I really began to form the opinion that Microstrategy may not be the best tool available for serious BI work. It's possible that more time behind the wheel with it would make it seem more powerful to me, or it may be possible that the commercial version is more comprehensive. I formed this opinion, because I found the need to export data into Excel to do any analysis beyond reporting. This didn't seem so bad-- after all there was a tutorial on precisely this feature.
I imagine that someone who has been doing this for a while and that has experience with Microstrategy may be able to do some of the more hard-core analysis in the tool. I am not one of these folks, thus I used the Excel standby.
The weaknesses that I refer to are those that would help someone identify trends across data using the tool. I applied filters and ranges, etc. this was great, but it is really only beefing up the "WHERE" and "HAVING" criterion conceptually. What I was really hoping to see was a tool that could see and point out that there may be a trend in revenue amongst 41 to 60 year-old females on its own. Or how about that the state of Washington appeared to not be on the radar much at all in 2007, while ramping up revenue over 2008 and 2009.
I was able to notice a couple of these things by myself using Excel. However, I did not find this type of thing to be an obvious feature of Microstrategy.
Beyond the tool, I imaging that Cognos or Tableau may have some of these power features built in. I could be wrong too, but I hope not. Anyway, regardless of the tools, it is clear as to why we currently have such a buzz around BI and data-scientists. This is real work using sophisticated tools that require experience and a knowledge of how to present meaningful statistics derived from large amounts of real data. In addition, one would also need to have a decent amount of domain knowledge about the organization where the data is being used.
I work in a university setting where it is all about enrollment, potential enrollment, retention, graduation rates, etc. It doesn't sound all that difficult, but across multiple campuses in a system, etc.-it does require some knowledge of the paradigm, the nomenclature/terminology and the data itself. It's obvious that it differs quite a bit from revenue, etc. However, the need to forecast is equally important if not as equally hard to predict accurately.
In closing, the last couple of weeks were useful introductions in the rabbit-hole of business intelligence. I hope that I grasped some of the concepts well enough to use them effectively in my career and that I didn't totally miss the boat on trending and forecasting in Microstrategy. Hans Rosling made me feel as though I missed something completely.
No comments:
Post a Comment