Skip to content

Data science consultancy in Zwolle

We build models that score and compare, so your team knows where to look first. That’s how Van Amersfoort Racing tracks young drivers.

With Van Amersfoort Racing, we're developing a tool that lets the racing team track and compare young drivers

Finding talented drivers is a challenge: just as on the track, you want to stay one step ahead of the competition. Together, we're developing a smart tool that makes it easy for Van Amersfoort Racing to track and compare young drivers.

Read the case study

Which request, client or candidate should we look at first today?

More cases, requests or signals come your way than your team can review thoroughly. A model scores and ranks them, putting the most important first. Your team makes the decisions, but starts in the right place.

Book a meeting

Score, rank and compare, with an explanation for every position on the list

Discuss your challenge

Comparing talent

For Van Amersfoort Racing, we’re developing a tool to track and compare young drivers.

Faster insights for agents

For a sports management agency, we built a platform that gives agents faster access to relevant player information.

Customers at risk

For DPG Media, we worked to reduce churn through advanced data analysis: who is likely to cancel, and who should you contact first?

Explaining the score

Each position on the list shows which factors carried the most weight, so an employee can check the calculation behind the choice.

A ranking is only useful if your team trusts it

  1. What makes a good choice

    Agree together on what would count as the right choice in hindsight, and which data could have predicted it.

  2. The model alongside the expert

    Compare the model’s ranking with your team’s ranking for cases where the outcome is known.

  3. In the workflow

    The score appears where the team already works, with an explanation and a regular check that it remains accurate.

Do we have enough data to score cases?

Talk it through with us

A good fit if

You handle many cases, you know afterwards which choices were right, and the outcomes are recorded somewhere. A model can then learn what makes a good candidate or application.

Not a good fit if

You handle only a handful of cases a year, or nobody records what happened. In that situation, a standard checklist or a good report is the better first step.

What determines the scope

How many sources are involved, whether outcomes are already recorded, how the score fits into the work and who will maintain the model afterwards.

Does the model make the decision?

No. It ranks cases and shows why. The decision stays with your people, and where personal data is involved, we agree in advance what the model may use.

Fixed rules, a simple score or a model that learns: which fits your decision?

Fixed rules

Choose this if a few clear criteria determine the decision. Rules are easy to explain and maintain.

A simple score

Choose this if you want to weigh a handful of features. Our advice: start here, then compare the result with what your experts choose.

A model that learns

Choose this if there are many cases and features, and the outcomes are recorded. The model can then learn patterns that a rule would miss.

Since 2016, we’ve been building models for sport, media and organisations beyond Amsterdam

Data Science Lab has been around since 2016 and is a Microsoft and Databricks partner. We’ve worked with Van Amersfoort Racing and DPG Media, and share what we learn at our Techdays.

Book an introductory meeting

The decision your team makes every day is what we’ll discuss in our first conversation

Stef Stoelwinder

Stef Stoelwinder

Business Development

Stef is your first point of contact. In our first conversation, we’ll discuss the decision your team makes every day and what data is available.

Which batch of applications, case files or candidates would you like to rank?

Tell us what decision your team makes and what data is available. We’ll reply by email or arrange an initial conversation.

Enter a valid email address.

Let us know how we can help you.

Frequently asked questions

It builds models that predict, classify or rank, and makes sure they are used in day-to-day work. This page focuses on scoring and comparing: which application, customer or candidate should receive attention first.

You can find other applications under AI applications.

No. We work from Westerdoksdijk in Amsterdam. We work online with organisations in Zwolle and the rest of Overijssel, and visit to test the results with your experts.

None of the case studies here took place in Zwolle; they show what we build.

By comparing them using cases where the outcome is known. You can see where the model and your experts agree, and where they differ. We discuss those differences in particular, because they show what the model does and does not know.

The model is only put into use once it ranks at least as well as your experts.

Often classification or regression models that produce a probability or score; sometimes a ranking model. The right choice depends on the data and how explainable the result needs to be.

You can read more about the techniques under machine learning.

By deciding in advance which characteristics the model must not use, then checking whether the results differ between groups. When people are involved, a DPIA is often needed.

You can find out what we provide for this under DPIA for AI.

Yes, that is a common application. We worked with DPG Media to reduce churn through advanced data analysis.

You can read how such a model is built under predicting churn.

Where the team already works: in the CRM, a worklist or a report, along with an explanation. A score displayed on a separate screen usually goes unused.

You can read how we keep a model running in production under MLOps.

When there are too few cases, or when nobody records the outcome of a decision. A model cannot learn in those circumstances, so a checklist or better record-keeping is needed first.

We also advise against a model if a few clear rules already lead to good decisions.