Your Code Doesn't Matter

Here Is What Actually Does

"My code is terrible."

That's what one of the best data scientists I’ve ever worked with used to moan about. He's been a data scientist for years, and he was genuinely concerned that his Python wasn't "elegant" enough.

The irony is his models had saved the company tens of thousands in operational costs annually.

I've seen his code and sure, there are redundancies here, inefficiencies there and he doesn't always adhere to clean code principles.

Here's the thing.

This conversation reminded me of a painful truth in our field: We're optimising the wrong things.

The Three Paths

There's a pattern I've noticed after years in this field, talking to hundreds of data scientists, from fresh graduates to tech leads:

Average data scientists optimise code.

Good data scientists optimise impact.

Elite data scientists optimise decisions.

Let that sink in for a moment.

The average data scientist obsesses over clean code, perfect documentation, and the most efficient algorithms. They spend hours refactoring code that already works. They lose sleep over whether their function names are descriptive enough.

Don't get me wrong - clean code matters. But it's not what makes you irreplaceable.

Good data scientists understand this. They've made the mental shift from "How can I write better code?" to "How can I create more value?" They optimise for business impact, not GitHub stars.

But the best data scientists I’ve met - the ones I strive to be like - they're playing a different game entirely.

The Elite Mindset

Elite data scientists understand something crucial: Data science isn't about code, or even impact - it's about decisions.

They ask questions like:

- Will this analysis change how the business operates?

- Is this the decision that needs to be made right now?

- What's the cost of being wrong vs the cost of being late?

They might write "messy" code that delivers a critical insight today rather than perfect code that arrives too late to matter.

To put it simply, they are fundamentally looking to validate or alter a company’s strategic decisions.

The Complexity Trap

Junior data scientists use complex solutions to show off their skills. It comes from a subtle insecurity of wanting to prove that you're good enough and that you do know your stuff. Senior data scientists use simple solutions to show off their understanding. True expertise isn't about what you can add - it's about what you dare to remove.

I’ll tell you one of my embarrassing stories to emphasise this.

It was my first year as a data scientist and I had finished sweating over my first predictive model that I thought would have an impact in the real world, not just to pass a module. I'd used three different models, ensemble methods, complex feature engineering - the works (my professor would be so proud).

I took it to my line manager - a data scientist of years.

Long story short; he replaced it all with a simple linear regression.

It performed 4% better, was easier to maintain and explain and most importantly it only took him the afternoon to build.

Not gonna lie, I'm still annoyed cause I spent weeks on my model but something clicked.

When I built my model,

I wasn't showing expertise. I was showing off.

Choosing Your Path

So, where are you on this journey?

If you're spending more time worrying about your code quality than the decisions it influences, you might be stuck on the average path.

This is a battle I am still fighting every day.

If you're measuring your success by impact but not yet thinking about the decision-making process, you're on the good path.

But if you want to be elite? Start optimising for decisions.

Ask yourself:

- Am I building models that make decisions easier?

- Do business leaders come to me before making crucial choices?

- Are my insights changing how people think, not just what they know?

The Way Forward

The path to elite data science isn't through more coding tutorials. It's through understanding how decisions are made in your organisation.

Start small:

1. Before your next project, ask "What decision will this influence?"

2. When presenting results, lead with the decision, support with the data

3. Build relationships with decision-makers - understand their thought process

The code will follow. The impact will come. But the decisions - that's where the real value lies.

Leave a comment below: Which path are you on? What's holding you back from the next level?

Your data science buddy,

Nash Jay M

P.S. Hit me up on Instagram if you want to discuss this more. I love hearing your thoughts.