The Appliance of Data Science

The farmer knows how to plough the field and has little interest in the fate of Icarus. The painting shows parallel worlds where the farming continues oblivious of the hardware testing in the bay. These parallel worlds exist within corporations, between business and technology. An all-too-common situation in oil and gas. How does a business ensure that Icarus and the Farmer work together rather than continue in isolation?

A major challenge in applying Big Data or Data Science is that the case studies tend to focus on the big scale solutions – billions of transactions, bucket-loads of data analysed, warehouses of servers (the cloud in a shed perhaps). After all this is how the technology and concept of big data was established. This kind of investment may be required for large-scale solutions, where there are larger scale benefits but simply buying the technology doesn't solve anything. There are big data opportunities in every direction in the lower cost, higher efficiency world but the organisation needs to know how to look.

The Appliance of Data Science should never start with buying technology. The journey should start with the business problem. Developing a repeatable method, along the way identifying where and what kind of technology is needed ensures that investment and effort is targeted. Working this way develops the company’s capability. Working closely with the business begins the process of change.

Not a Software Tool, a Method

“ What software did you use for the ‘Shows’ mapping in the North Sea?” I was asked by one of a newly assembled Data Science team in a very large European energy company. I could simply provide the shopping list of tools that were used, but for two things. Firstly, the software used was generic, useful at this point, cheap and easy to install, but probably not suitable for any kind of larger-scale implementation. Secondly, and more importantly, a focus on software first is always a mistake. The secret of our approach is to be confident in pushing back on the stupendously well-funded hype around big data tools and to focus on solving business problems. 

Appliance Method Principles

The principles of this methodology are as follows.

1. Start with the business decision

2. Focus on the data used in that decision

3. 'How can improving the data improve the decision?'

4. Business experts identify value 

Bruegel saw that Icarus was dreaming of solving a technical challenge. He saw that the Farmer is focused on the task in hand too busy delivering the crop. Within a company the technology has to be focused on reducing risks, reducing costs and helping finding and producing hydrocarbons. To ensure that Data Science effort is focused on business benefits the leadership needs to be engaged and drive the challenge. Buying software and then looking for the solution raises a barrier to the necessary cooperation, developing an appliance method and testing against real cases builds a bridge between the asset teams and the IT groups. Imagine Icarus and the farmer discussing how to improve crop yields - or perhaps the melting point of wax?