Ahead of his presentation at the Big Data Innovation Summit, I was lucky enough to talk data, his role at Gilt Groupe and the skills gap with Igor Elbert.
Igor Elbert is a Principal Data Scientist at Gilt, where he applies machine-learning techniques to study Gilt’s unique customer base and add value to partner brands. Prior to joining Gilt, Igor worked as a Vice President of Quantitative Analytics for Barnes & Noble, where he oversaw deep analytics on loyalty and customer behavior. From calculating financial risk for Salomon Brothers to tracking movements of millions of items across the supply chain for major brands, Igor has been pushing innovative data analysis to new frontiers for more than 20 years. His current passion is a guided discovery.
Gilt Groupe's unique customer base has created a fascinating data challenge for Igor, what with the uncommon aspect of the products that are included on the e-commerce site. The products are generally of limited availability of one or two days. This means that one of the biggest challenges is creating metrics to link products, what makes somebody like one pair of shoes and not another?
This kind of unique data challenge has added a degree of difficulty in recruitment for the analytics department at a time when we are seeing a significant issue in the numbers of adequately qualified data scientists across the entire big data community.
I asked Igor his thoughts on why there was this skills gap.
His response was that there are essentially two competing mindsets that make the best data scientists and finding people who have both is difficult. On the one hand there needs to be a technology knowledge and mathematical skill to utilise algorithms, but then there also needs to be business knowledge and savvy in order to answer business questions within the data. Finding one or the other can be much easier, but finding both is difficult, especially when most analytics departments are looking for the same thing.
Igor is a man who knows what makes a great data scientist, having worked across a multitude of data projects from security aspects through to supply chain analytics. This has also given him a perspective that few have the experience to appreciate, on where big data may be effective next. Here Igor believes that guided discovery is an area that will be affected by the influence of big data.
His thoughts are that with the use of guided discovery it is likely that big data and data science as a whole will not be reliant purely on the strength of an algorithm, but will be able to interact and change based on the end user interaction.This could be anything from the way a customer interacts with a display to the way in which a data scientist is attempting to manipulate the data.
This also matches up with Igor's thoughts on where data will go within the next 5 years. His belief, which is shared by others at Gilt Groupe, is that within a year what is considered as big data today will simply be data. We will see what we consider to be large datasets today, becoming the norm for companies within the next few years.
The challenge will not be the management of data this size, but the ways in which to communicate the findings from the data in an effective way to make the most of it from a business perspective.
Igor will be presenting at the Big Data Innovation Summit in Boston, September 12 & 13 2013.