This is my first post with SCN. I am not an SAP user, I found myself in this ecosystem in 2009 when DataXstream LLC, an SAP Partner, needed marketing assistance. But our focus is on finding value add for the client, both in our staffing/solutions offerings and in our marketing efforts and so I am included in the team to attend SAP TechEd, Las Vegas.
I found Nate Silver’s keynote last night quite compelling. We have clients we are working with currently who are dealing with the very issue of data analysis. And they are insisting on using 3rd party software with algorithms to crunch their data and predict purchasing materials. When asked how will they test these results, they look at us blankly. After the keynote we met with a friend who experiences this in his place of work. The company demands real time data, caught in the hype of real time decisions, but isn’t able to analyze the data to make those decisions.
When SAP released HANA the analysts all replied, “big data, faster, great, but what will you do with it?” And there is the potential to make bad decisions faster, Mr. Silver illustrated this with an example of an erroneous tweet by a news agency that resulted in the stock market rapidly dropping in value and then recovering almost as quickly.
Silver spoke quite a bit about Bayesian Statistics, and the need to think in probabilities, the need to experiment, the need to err and learn. And he ended with Peit Heim’s “Road to Wisdom” Err and err and err again, but less and less and less.
The question for me now is, how much can a client afford to experiment and err? In today’s business world there can be serious consequences to bad decisions. This points to the need for data scientists. And brings me back to the example of the client who has chosen to use a third party software to analyze the data for them, essentially provided them with someone to blame for mistakes, without taking into consideration their need to test the results, or fine tune the results.
In the hype surrounding more data faster, the need for data analysis is easy to overlook. Mr. Silver cautions, there is a need to apply human skepticism to the data you produce.
I do not know if this keynote will be available in the replays, if it is and you missed it I highly recommend it.