MinneAnalytics is a local (Minnesota – Twin Cities) non-profit organization dedicated to serving Minnesota’s Data Science and Analytics community. As the saying goes – For more information about MinneAnalytics, check out their web site at http://minneanalytics.org or on the Twitter (@MinneAnalytics).
FARCON is a Financial and Retail industry focused conference for latest information in Data Science and Analytics in that space. FARCON is an annual conference, and 2017 was the third annual conference. MinneAnalytics has partnered with University of Minnesota Carlson School of Management for all three FARCON conferences.
FARCON 2017 Location
University of Minnesota Twin Cities campus has hosted all three FARCON conferences. Because of the number of participates in this year’s conference, FARCON was conducted in the beautiful, new McNamara Alumni Center building. For me, it was great to get back to U of M campus – lots of new and remodeled buildings since my undergraduate courses many moons ago. The thought I would look older than U of M campus buildings never crossed my mind twenty some odd years ago when I went to freshman orientation. But I can say now that ship has sailed.
A few general, overall comments about FARCON 2017
1. Agile Methodology in Data Science Projects
Nothing adds project cost faster than rework. Rework will cost both in terms of time delays and dollar costs. This is true across any industry – IT, constructing a building, designing a new airplane and so on. For Data Science projects – Involving business partners early and often is very helpful in obtaining overall project success. Using Agile’s model of frequent meetings between customers and IT help Data Scientists gain better understanding of business. The frequent engagement of customers builds confidence in IT by seeing regular project updates and re-prioritize objectives if necessary. “We should all swim in the Data Lake together” will become a new theme in the Data Science community.
2. Venn Diagrams can be very descriptive
I was fortunate to select Mohammed Aaser’s presentation “Over Half of Analytics Fail: What separates those that succeed?” for the first break out session of the day. Mr. Aaser provided a good description of a Venn diagram, and a couple other presenters later in the day reference this slide. A simple Venn diagram can visually communicate a lot of information to coworkers and business partners.
3. DevOps Commitment by Organizations
I was very impressed by the number of companies (small and large) that have adopted DevOps to some degree in their organization. Changing a business organization to a DevOps model is a very large commitment and it was nice to hear it has been done as much as it is.
4. The future is now – play to win
Many Data Science projects begin with broad, ambitious goals without really understanding key objectives. Project can start with a set of assumptions of availability and/or frequency of data, some of which may turn out to be wrong. Work to direct Data Science project to succeed rather than trying to avoid failure.
5. I need a new laptop 😉
I picked up a couple new stickers at FARCON. These new stickers look very much like the old stickers on my current laptop. But these are new stickers, so I’m sure they are better than the old ones I have. And since I’m out of “sticker real estate” on my current laptop… Alright, lack of “sticker real estate” on a laptop may be a first world problem when justifying a purchase of a new laptop, but I’ll stick to it for now…
Quick Summary of FARCON 2017 Sessions I Attended
General Disclaimer: The following summaries are just one or two aspects of an overall presentation that I attended.
Over half of Analytics Project Fail: What Separates Those That Succeed
I mentioned earlier the Venn Diagram example. Mr. Aaser used a Venn Diagram to illustrate the importance of interactions between Analytics, Technology and Customer teams. Mr. Aaser also defined a need of a “Translator” that is very important in Data Science projects. A “Translator” is someone that interacts with the three teams to ensure the right communication is taking place. A “Translator” would be more of a jack of all trades type of individual, and may not be deep into code or business requirements. But the “Translators” need to have strong problem solving skills and drive business solutions.
Better Together: Using Qual and Quant to Deliver a Better Customer Experience
Angelica Bonacci, Jim Tincher
Customer journey mapping is the end-to-end experience from customer point of view. To evaluate customer experience – look at overall customer journey instead of individual touch points. In business and IT, processes develop over time. If there is any negative customer feedback, we tend to think it was because of one aspect of the process and want to focus a fix on that one thing. By looking at an overall customer journey, we gain understanding of what a customer goes through from start to finish of a business process. This experience can shed light on fixing a root cause of problem instead of focusing on small symptoms that do not improve overall customer experience.
Executive Panel: Analytics, Big Data and Artificial Intelligence Impact on Retail and Financial Industries from the Executive Perspective
Moderator: Ali Wing
Speakers: Mohammed Aaser, Eric Bibelnieks, Julie Joyce
Interesting questions regarding how Big Data is changing companies; what tools companies are using; complexities of legacy systems and the tradeoff of agility and trust. I also heard a good explanation of a Data Lake turning into a Data Toxic Swamp. I didn’t know there was a non-four letter word term for that environment.
Performing Analytics with the Data You Have, Not the Data You Wish You Had
Mr. Cooley made the case as Data Scientists we are responsible for finding solutions that work today and solving problems now. There will always be the situation where “better data in the future”, but our customers can’t wait for that. What questions can be answered by the data we have now?
Counterintuitive Tips to Conquer the Productivity Paradox in Analytics
Mr. McNellis had a few recommendations in this presentation. The first, stop focusing on strategy and focus on execution falls in line with Agile methodology. We need to measure execution compliance and segment results for execution opportunities or best practices. I also liked Mr. McNellis’ reason behind “Partner with Losers” in an organization since they have the most to gain from a Data Analytics project.
When a “What If” Becomes a Reality
Scott Ernst, Ph. D.
This was an interesting session on setting up DevOps organization in a startup company. Dr. Ernst provided a good description of Data Lakes (compute and storage) and Data Governance suggestions to avoid a Data Swamp (keep data immutable in a Data Mart). I liked the Service Model of “Pets” and “Cattle”. “Pets” applications are where IT goes through great pains to keep a server or application alive. “Cattle” applications are situations like a Docker container model where if performance goes bad, fire up a new instance and kill off the “problem” instance.
Community Banking Analysis
I had to step out for a call and missed the second half of this presentation. For the first half of the session, Mr. Richter provided a case study of a small, family owner Community Bank transformation from “hand shack” deals to a cold, impersonal, modern banking business model. (I may be paraphrasing a little bit…) Seriously it was interesting why this company was brought in and what data they had to work with. Over one or two years, the consulting company had to come up with recommendations for strategic decisions. This was done in a phased approach. First phase (Data Foundation) started with defining repeatable processes and maintaining data compliance. Second phase (Understand Data) included creating customer profiles, cost models and visualization results. Third phase was Enhancing Customer Profiles. Final phase (Make Strategic Decisions) included reconfiguring product offerings and high value products.
The Importance of Data Strategy and Governance for Data Science Success
Mr. Chenard started the presentation with the Gartner Group survey that only 8% success rate for Data Science project. But at the same time, companies are increasing spending on Data Science projects. There were a number of terms that were defined during the session to help define Data Science projects and how to increase the success rate for developing a product. I’ll add a couple of the terms here. Traditional Data Governance tends to focus on compliance and often misses human activity. Data is a combination of people and actions and how do Data Scientists interpret that association?