Data Science Dinner November 5th 2015
On November 5th the first edition of the Data Science Dinner took place. The Data Science Dinner, an initiative of GoDataDriven, aims at bringing together executives who lead a data driven enterprise and/or are responsible for the transition to an organization where strategy and business decisions are based on data.
During the evening executives exchange ideas, best-practices and get a second opinion on their business ideas.
This time, Frank Derks, Head of Advanced Analytics at ING Commercial Banking, presented the 9 elements of a successful datadriven transition.
Over dinner, several datadriven topics were discussed including Industry 4.0, the ideal organization structure and experimentation.
Rob Dielemans kicked of by discussing the fourth industrial revolution. Traditionally, the elements of an industry where material, means and people. From mechanical (steam, water), industries started dividing labor and introduced mass production. Then automation entered the industry with IT and electronics. Now, data and connectivity are once more changing our way of working.
At the tables we discussed this, some of the outcomes where: - We used to have standardization to keep prices low. This is moving towards differentiation and personalization, where data makes it possible to keep costs low. - Data without context is useless. - Collecting data just for the sake of collecting can be seen as the pollution of Industry 4.0.
The ideal organization for the Data Driven Enterprise
Next up for the main course, Renald Buter discussed the ideal organization structure. Is this fixed like in a prison, hierarchical like in the army or loose like in a public space? We discussed the four aspects of organizational structure: Strongly versus weakly structured Disciplined versus loose interactions Hierarchical versus self-organised Monodisciplinary versus multidisciplinary
At the tables we discussed the ideal structure, and whether it is possible to create this ideal structure within a company or outside of the business.
Some attendees noticed that in their cases, the organizational change could only be made within the company, due to company size (small) or type of company (government body). Also, the support from the management is essential, if there is no vision from the management, chances are slim that an external structure will become a success.
Attracting the right people is crucial too. This is done by creating an attractive image as an employer by sharing a vision, creating an inspiring environment and projects that have a direct impact on the business.
When companies are able to attract the right people, they are able to create advantages of scale: 1 scientist would outperform dozens of analysts. So the right people also make it possible to operate in smaller teams.
Corporates are not able to experiment. Or are they?
Finally, as a dessert, Friso van Vollenhoven discussed experimentation within a corporate context. Companies say that they need to experiment, but are companies capable of this or or does the nature of a business: the need to make money, limit the experimentation opportunities?
Friso’s statement that corporates are not able to experiment lead to a proper discussion about experimentation: is an A/B test an experiment? How do you innovate from an experiment? An experiment is a first step, and admitting that something is nog good or successful can be difficult, although to tell this often is more difficult than to listening to it being told, as long as there are also successful experiments. Support from the management is good, but when experimenting, involvement from the management is not desirable as it will contaminate the experiment (“can we tell upfront which part of the A/B test will be the winner?”)
A culture where failure is accepted leads to higher achieving businesses, but companies need to make money too; you need successful experiments to make up for the unsuccessful ones.
The conclusion was that companies are very well capable of experimenting, perhaps the pressure from the business leads to an optimal experimentation culture: where many experiments can be done on a small scale, eliminating the unsuccessful ones and enabling the successful ones to blossom.
Interested in joining the next Data Science Dinner, scheduled for Spring 2016? Contact us by pinging us an email email@example.com
Follow us for more of this
Deep Learning Blimp
October 16, 2018
Big Data Expo 2018: Deep Learning, the Engine of the AI Revolution
October 05, 2018
How Blockchain Technology Will Make Platforms Obsolete
September 27, 2018
Wasting money with data science
September 21, 2018
How to Find Blockchain Use Cases: Part I
September 17, 2018
Opening up some training material
September 05, 2018