You’re already using optimisation every day as a data scientist - but you probably don’t think about it.

Every time you train a machine learning model, you’re running an optimisation algorithm to find the best parameters.

But there’s a second type of optimisation that most data scientists never even touch - and it’s where the real business value often lives.

It’s called decision optimisation, and it can transform your ML predictions into actionable decisions.

Here’s the difference:

• ML parameter optimisation is about finding the best weights for your neural network so it makes accurate predictions.
• Decision optimisation takes those predictions and uses them to determine the optimal production schedule, delivery routes, or inventory levels that maximise profit while respecting real-world constraints.

One predicts the future. The other tells you what to do about it.

In the latest episode of Value Driven Data Science, Dr Tim Varelmann, founder of Bluebird Optimization, joins me to explain how combining these two types of optimisation creates solutions that go far beyond prediction - helping stakeholders make better decisions in uncertain environments.

You’ll discover:

1. How decision optimisation differs from ML parameter tuning [02:19]
2. Why combining predictions with optimisation multiplies value [13:36]
3. The mindset shift needed to think in optimisation terms [22:59]
4. How to spot immediate optimisation opportunities in your work [23:42]

Most data scientists stop at prediction. The ones creating real impact connect those predictions to the decisions that matter.

Listen now on Apple Podcasts or Spotify, or click the link below:

​Episode 96: Making Better Decisions with ML and Optimisation​

Talk again soon,

Dr Genevieve Hayes