UNLOCKING YOUR DATA FOR DECISIONS THAT MATTER

1

Introducing Nukon 2

Enterprise Architect

Ben Caldwell Gareth Williams

Our Capability

⚙️ Deployment of solutions 🔌 Connection of solutions

📝 Use Case Development 🛟 Use Case Support 📖 Use Case Coaching

🎖️ Training by certified trainers 🖥️ Integration into other systems ⏰ 24/7 Support Capability

Time Series Specialists

Seeq Specialists

Lessons Learned How to avoid the danger zone 4

Gareth Speako 4

Welcome to the Danger Zone! 5

Gareth What is the danger zone? Seeq has empowered a new wave of what we call Citizen Data Scientists — process engineers, operators, planners — who know the problem intimately, and now have the tools to explore their data in real-time, visually, and powerfully.” “These folks already have ‘hacking skills’ — they understand what to look for — and Seeq gives them the means to do it.” here’s the risk: “The accessibility of the tool doesn’t mean we’ve eliminated the need for rigour, validation, or understanding. And that’s where the Danger Zone begins.”

Easier to do data analysis but there are TRAPS to watch out for!

5

The Danger Zone Explained 6

Building great insights... with no path to operationalisation

Inadequate validation Misinterpreting correlations

Ben 6

What is the Danger Zone? 7 Danger Zone Current conditions make it easy to know enough to be dangerous Business data integration and availability Maybe you have a stats blind spot? Are you a data alchemist – looking to create gold from all the data lead you have access to? Have you gone down a data rabbit hole and always need ‘just a little bit more data’ to get to the answer?

Ben e.g. ChatGPT Low/no code analytics

7

No Hypothesis 8 Danger Zone We are doing this to increase our knowledge of the world There needs to be a motivating question that we will answer with data No clear hypothesis and masses of data at our fingertips is a perfect storm We end up data dredging and finding nonsense correlations.

Ben 8

Misinterpreting Correlations 9 Danger Zone

With enough data there will be spurious correlations Don’t start searching for answers in the data Form a hypothesis and select a small set of datapoints to test it.

Ben

All the data was pulled together in one place for the first time A super correlation finder algorithm developed The team determined that UFOs exist... and to everyone’s surprise they run on kerosene

9

Inadequate Validation 10 Danger Zone We create a model then validate to see how well it forecasts This looks pretty good! But if we zoom out... We haven’t understood the seasonality of our data.

Ben 10

Not Starting With “How Will It Be Used?” 11 Danger Zone To be successful a model or insight needs to change a user's behaviour Who is using it? What decisions are they making? When do they make the decision? The answers inform what you build Iterate with end users – try and get an MVP thin slice to them as soon as possible.

Ben This is not necessarily in the danger zone but it is a problem

11

Avoiding the Danger Zone 12 Know When You’re in the Danger Zone

If you're building models without validation… you're in it. If you're acting on trends without understanding the cause… you're in it. If you can’t explain your insight to someone else in the business… you’re probably in it. Validate With People Outside Your Skillset

The best insights come from collaboration; maths meets machines meets operations. Talk to your process engineer. Talk to your data person. Talk to your operator.

Gareth Start every analysis by asking: “What decision am I trying to support?” Build cross-functional review loops into your analysis.

12

Avoiding the Danger Zone 13 Invest in Repeatability, Not One-Off Heroics

It's tempting to build the “one killer dashboard” but what happens when it breaks? We see this all the time, great results that can’t be repeated or scaled. Use Seeq’s inbuilt tools such as the Journal Get Comfortable Asking for Help

Whether you're deep in machine learning or just dipping a toe in, we all hit walls. Nukon’s role is to bring together your cross-functional teams to generate outcomes.

Gareth Aim to build reusable logic, templates, and documented workflows. Use Seeq’s templates and Journals to your advantage. Ask yourself: “Would a 1-hour chat with a data scientist or process expert save me days of frustration or risky decisions?”

13

Nukon.com

.MsftOfcThm_Accent1_Fill_v2 { fill:#4EBC89; }