The Data Journey.
Data is important. Not cool, not fashionable, but important. Data, handled correctly, has the potential to transform business and revolutionise our lives. When you discover where data sits in the process taking considered action it becomes clear that data is a key resource that can be exploited to generate value. Having an understanding the of the journey data needs to take to support this value will give you a strategic advantage.
In his recent talk ‘What Is Intelligence’ Hermann Hauser used the Robotic description of intelligence ‘Knowing what to do next’ to explain how machine super intelligence is closer than we think. “The reason for this”, he said, was “the abundance of high quality data”. In this article I wish to show how the process of learning through data works.
The diagram shown at the start of this article is an evolution of the traditional ‘DIKW’ pyramid that is used in academia. I present the four elements of Data, Information, Knowledge and Action (DIKA). Wisdom, in DIKA, is part of the line between Knowledge and Action not a ‘pinnacle’ end goal. DIKA is a continuous cycle of learning from data driven by the desire.
Data is the exhaust of action. Any action, natural or artificial, when instrumented, generates a data exhaust. Data cannot come into existence any other way. Data in its raw form is bound to the ontology (https://en.wikipedia.org/wiki/Ontology) of the action. This ontology is often designed into the system that instrumented the action. Think ‘URL visit’ in web logs, ‘purchase record’ in ecommerce, ‘ticket gate open’ on public transport or ‘new geolocation’ on a wildlife tracking collar. Each of these ‘Data Points’ represent one instance of an action in their respective system. It is the desire for information that leads us to gather more data.
Information is always a derivation of data. It is a selection of some portion of the overall data set that has been extracted and often transformed in some way. Information can be found in many forms, from a report or table to an infographic or full blown interactive visualisation. An extremely important thing to remember about information is that the process of deriving it from data involves some form of bias. This can be as simple as selection bias (by wanting less than the whole you are leaving things out) to aggregation bias (removing time granularity can give a false sense of continuity) and bias inserted through manipulation (data smoothed to be easier to consume). Information is often mistaken for data in systems where real data is hard to get. For example, a report referred to as ‘the data’. As long as you are aware of this then information is valuable because it makes portions of data consumable and able to interact or combine more cleanly. It is the desire for knowledge that leads us to seek new information.
In the study of Knowledge (https://en.wikipedia.org/wiki/Epistemology) you will find a core part of the field working on how to define knowledge. Defining knowledge turns out to be very difficult on the grand scale. In the DIKA cycle, however, we define Knowledge as ‘that which grows through learning and drives action’. Learning is the act of combining pre-existing knowledge with new information. You will see the learning loop right at the center of the DIKA diagram. In discussions of epistemology there is a long running debate on how to separate Knowledge from Belief. It is possible to see a similar struggle in business. This struggle is often called ‘Wanting to be more data driven’. In DIKA without information derived from data, all ‘Knowledge’ is just ‘Belief’, an unsupported opinion often called ‘Gut Feel’. Actions taken on this basis are less accountable. An accountable process builds comfort and trust in decision making. A lack of accountability will erode trust in the decision making process. It is the desire for action that leads us to seek knowledge.
Action means a change in your environment. This could be a business changing a product’s colour or an individual purchasing that product. The knowledge learned from information derived from data doesn’t itself doesn’t set the action in motion. The knowledge gained leads to more confidence in a decision to take action. In DIKW this can be what Wisdom is for. It is possible to add an extra layer of granularity to DIKA to show this. Knowledge makes the suggestion of the action easier and leads to greater trust that the action is justified. For example, after studying information derived from advertising spend data, an agency decides to increase spend on a particular advertising keyword. Spending is the action. This action has to be justified to the client. Justifying this action, and maintaining client trust is easier when the knowledge used to make the decision is learned from information derived from data. As we saw in the Data section. An action instrumented then generates more data and the cycle may continue.
You will see Desire at the heart of the DIKA cycle working counter clockwise. Desire is what inspires activity at each step. In the same way that desiring to take action leads you to seek knowledge, you must desire information in pursuit of new knowledge. This can be seen very clearly in sentences commonly heard around business. ‘We need to get more users to our website? I wish I knew how to do that.’ There is an obvious desire to take action here. It is also obvious that the person speaking implicitly understands that that action must be driven by knowledge. There is an interesting parallel case, that shows the same mechanism in action. The case of recommendations on a website. The owner of the website desires customer action; a purchase. If a user knows about relevant products, they are more likely to buy them. They provide a user with more information (the recommendations) designing that this information will combine with the user’s existing knowledge to result in the action of purchase.
Throughout I have described DIKA in terms of people. However, it is quite possible to consider the process of Machine Learning (ML) to follow the same course. An ML Model is trained on information derived from data and represents knowledge used to support a machines’ decision to take action. The desire is still only human.
To Conclude, the DIKA cycle, otherwise known as ‘The Data Journey’ evolves the idea of the DIKW pyramid by recognising that action is more valuable to business than the academic pursuit of wisdom. It also gives context to data work explaining why this work should be done and showing how this work is directly tied to value. As the cycle turns, data leads to learning, which leads to action. Action generates value … and more data.Published in