Time series data model

Our mission is that our clients can act data-driven in their daily operations. But we want to go further — we want to empower them so that they:

1. Discover value in new information streams
2. Understand how uncertain information leads to acceptable risks
3. Streamline the innovation process

From our professional experience (and in collaboration with CWI), we developed our Time Series Data Model. It consists of a suite of open-source Python packages for predictive analytics that serve to take away barriers between IT developers, data scientists and domain experts (energy, agriculture, etc.).

Powers

I’ve discussed the three powers, which we made it part of our mission for our clients to achieve: Value Discovery, Risk Weighing, Ownership. Why did we choose these three?

We believe that technology is useless without people directing it. Digital technology can help discover great customer value, but your team still has to make the discoveries. Digital technology is not only powerful, but also expensive, so when using it, one needs to weigh the risks of wrong timing or wrong directions. And your management team needs to make the final decision where to go — decisions should be made by someone who experiences ownership. Our clients and Seita will discuss the options and do the risk weighing together, but the client is the owner, and they should feel that way.

Segments

Powers don’t come about by saying their names. We thought long and hard about what underlies these kinds of successes, based on our professional experiences. The three segments which we know are crucial for a successful data-driven organization are:

1. Data
2. Team
3. Product

These segments and our three powers are not related one-by-one. Instead, each power is achieved when two underlying segments “click”. The graphic shows that nicely. Let’s go around:

Value discovery requires that the right data is known and that the product process is capable of bringing that data into features. Managing risk also requires a product process which is capable of finding out what does not work (and stop doing that), while building a team the right way plays a huge factor in whether this is possible and also is a huge cost center in itself. Having ownership of the process is certainly hard if there is no technical person on staff (if, say, all technical expertise is by consultants or out-sourced) — a healthy mix is probably reasonable. Finally, data also plays a huge part in having ownership. If all that data is governed by service providers (for example, all your dashboards are made and hosted by some third party), you might not feel empowered enough to succeed.

The trick is to find out exactly how all this applies for a specific case — your case. It pays off to think a little deeper into segments:

At this stage, our thinking is driven by two questions: In the theme of acting data-driven in their operations, which goals should we set for a client in each segment? And which capabilities let our clients accomplish these goals?

Segment goals

Let’s first list the goals. We’ll explain why we formulated each goal and how our advice usually looks like. Each goal is combined with a constraint (decisions, time, costs), which should force us to think of the context for which these goals are crucial.

Segment capabilities

To be able to concretely spell out what each segment might be accomplishing in a data-driven state, we formulated capabilites per segment. Each capability adds huge value and needs a plan, concerning who, why and when.

We are working with this model in our Technology Roadmap Workshop.