Skip navigation

AWS Cloud Drives Amazon's Q2 Revenue Growth



Amazon Web Services -- which is gaining more big data capabilities -- accelerated the pace of its cloud growth, generating $1.82 billion in second-quarter sales.


Much was said in broadcast media yesterday about Amazon surpassing Wal-Mart as the largest retail provider in the U.S. What has not be discussed as much, except in IT media, is Amazon's drive to the forefront of cloud based services. Not only leaving tech giants like Google and Microsoft in the dust, but doing so at a phenomenal growth rate.


To put this in perspective, any Joe with a computer, internet access, and a credit card can have Amazon host everything which would previously require your own IT department: web servers, network databases, application software. Such capabilities at such a cost savings was unheard of just a few years ago. Of course, we could also point to Tableau having it's own clouding hosting services for viz displays and data.


So what does this portend for the future of data analysts? Here are a few thoughts:


  • This does not address access to data sources, but that’s rapidly evolving as well. I wouldn’t be surprised if Jeff Bezos and company realize they could also make money selling subscription data.
  • Such tools will help support more free agent data analysts who don’t need to be tied to organizations for IT support.
  • Talented individuals and small groups will have tremendous agility to deliver analysis services on par with larger more established organizations. Also not be constrained in the scope of what they can do by the organization’s priorities and resource limitations. Cloud services help level the playing field.
  • Not a warm-fuzzy feeling for IT groups, but executive leadership within organizations looking to outsource related services to support their staff of data analysts at a lower cost.

My apologies in advance for the long dissertation. I’ve been thinking of formats to mix things up at our user group meetings, particularly in ways which will benefit new users or those interested in seeing what Tableau can do. But also challenge the creativity of our more experienced users.


Tableau has great events like the Iron Viz Contest, where users compete against the clock to design the most creative and informative displays. Sure to generate a-ha’s with people well versed in Tableau, but risks deer-in-truck-headlights reactions from those who have a minimal understanding of the product. With the release of 9.0, Tableau emphasized the theme of Analytics in the Flow which is an important part of the flexibility provided by Tableau. But this presupposes you have the data you want and some idea of what you want to do with it. As with any data analysis, it begins with the data and some sense of strategy for how you want to approach the analysis - well before you make your first click in Tableau. We also need to slow things down for new users so they have an idea how to make this transition from examining data content and strategy to getting in the flow with Tableau.


Years ago I attended a competitive intelligence training program where the instructor made a statement which has always stuck with me: “Good analysis begins with good questions. What are the questions your decision makers need answered?” For our August 21 session I would like to begin there with an example set of data (attached). This was created from publicly available file on the Texas Department of Health Services website which I modified and formatted for this exercise. It gives basic location information of inpatient facilities in Texas along with their licensed bed count by bed category. After we cover any UTMB Tableau related news, we will devote the rest of our time to a three part “Data Discovery Session”.


Part 1: Good analysis begins with good questions.


In my opinion, the Zen of data analysis is the ability to see into a set of data and grasp the potential of what you can learn from it. In advance of our session I would like you to look over the first worksheet tab and think about questions we can explore with the content available in this file. For example, from a population health standpoint:


  • How disproportionate are the availabilities of inpatient beds in urban versus rural areas?
  • Lack of mental health facilities is a major concern. What are the ratios of Psychiatric to other bed types?
  • Considering the aging population, distribution of rehab facilities and capacity will be a growing concern. What does this currently look like?
  • Facilities renovate over time, but for purposes of this exercise the “License Effective Date” represents the date a facility was originally licensed. What can we tell about the aging of our inpatient facilities vs. new ones coming into operation?


These are just a few ideas and I’m hoping you can come up with other even more creative angles we can explore. We will spend a few minutes at our meeting brainstorming these ideas.


Part 2: Experiment with Tableau displays to address the questions


For the experienced users also think about potential Tableau displays to address your questions. After we finish the question brainstorming we will start from scratch building Tableau displays in a manner where we can walk new users through the process. We will be open to trial and error (trial and error in the sense of trying a viz to see if it tells us anything useful and just go back to building another if it doesn’t). Then come up with a set of worksheet we want to use.


Part 3: (time permitting) Assemble worksheets to dashboards.


A few of technical points about the Excel file:


  • The first worksheet tab gives you a better view of the content in a flat table arrangement, but does not work best in Tableau. I transposed the data using Tableau’s Excel Data Shaping plug-in to a format more workable with Tableau. So if you want to play with the file in Tableau prior to the meeting, make your data connection to the “TXHosp2015_05-Tableau” worksheet tab.
  • The counts are for licensed beds. We realize actual operating beds in a facility could be fewer. But we will go with that for demonstration purposes.
  • UTMB Galveston and MD Anderson are missing from this table. They are considered state hospitals, are exempt from licensing, and therefore do not appear on this list. Again just for demonstration purposes we will go with facilities in this file. Angleton Danbury had a preexisting license prior to merging with UTMB last fall, so that is why it appears on the list.


So I hope you agree this will be a more interesting and engaging exercise for our next meeting. Hope to see you there.