Working with Big Data requires an agile approach

In a packed conference center in Amsterdam, many people from industry came together to listen to speakers during the Big Data Analytics Europe event. An event was organized by Whitehall media.
Authors of the book “The Chief Data Officer’s Playbook” Peter Jackson and Caroline Carruthers told the following. “Many companies do not have Chief Data Officers (CDO) and when they appoint someone they think that all data issues are drawn up in one go”. It is the CDO with relevant team members (and significant input from key stakeholders) that will enable them to deliver maximum value by transforming the raw data into metadata (or smart data). This can only be successful, as a first starting point, if the company asks the CDO the right questions. CDO can only be successful if the (whole) organization takes part and becomes part of the journey.

KLM

Cosmando Byarugaba, Program Director, Big Data Program of Air France / KLM: “Building blocks with working with Big Data in combination with AI consists of a vision, long-term strategy. If your manager asks you: what and how much money will it yield? Then that is a wrong question. We only know that when we have been busy with it. Companies must believe in their employees and also give them the chance to experiment. Support from your managers is very important in this. This was immediately the message of all companies: work agile, lean and experiment fully. Alejandro Leon, director of Lead architecture digital at Philips IT, also said the same thing. Philips IT works with the principle: fail fast, react, learn and change. They want to shorten the release cycle.

 agile-works-2
Cosmando Byarugaba: Team of Data scientists and Data engineers of KLM worked on textmining, clustering of algorithms and they could predict 95% or a potential customer on the website a “business” or a “leisure” customer.
 agile-work-1

KNMI

One of the speakers was Michiel van Dongen (Business Developer for R & D and Innovation) at KNMI (Royal Dutch Meteorological Institute) . KNMI has been a data factory since it was founded in 1854. KNMI has been collecting data for years via Observation networks, which they have from the beginning and satellites. They also increasingly focus on what they call “non-traditional data”. These are data that they retrieve via:

  • telecommunications towers
  • smartphones
  • aircraft
  • vehicles (cars)
  • cameras

Using an example, they show how they use models to predict risk (dangers). They have a system consisting of sensors on a road, but since last year they also tested with car data. More and more cars are collecting data. They link different data points on the road, what they observe via cameras, other sensors and then they can make accurate decisions about eg warnings. They also carried out tests with sensors on bicycles.

 agile-works-4
Another example is how they measure an ice build-up in the atmosphere. Only ice build up in the atmosphere does not immediately show what effect it has on the ground. Is there ice on the ground? They now use insurance claims to validate this. They have to do the correct forecast (prediction). If they too often predict what is wrong, this can have the opposite effect whereby people no longer take the warnings seriously.
They also measure solar panel data (which is used to generate energy). It is very important for KNMI to have long-term data sets for scenario planning. They have the following R & D issues that they can answer with this, such as how often does a weather condition happen? how long does it take? Etc. KNMI works with many parties and they are looking for more parties.
The most important take away of event was: “Do not ask what i twill earn, you have to believe in it” and actually “go and try, experiment”
 
Loading...