Notes / September 2026

When a simpler method was enough

A transformer where logistic regression did the job, and a pipeline that Excel already covered.

1 minute read.

The transformer

I once presented a complicated transformer architecture. My instinct was that a more sophisticated model had to give us something better.

What I tried
Logistic regression, on the same task.
What surprised me
It did the job. And it was easier to understand, and to explain to the people who would use it.
What I did next
We went with logistic regression. The model had to be deployed and used, not only shown in a conference paper.

The spreadsheet

Another time, I was explaining a pipeline to a team. Convert the data to Parquet, then process it through several stages. The team lead stopped me. Excel already did what the team needed.

What I proposedrawParquetstage 1stage 2resultWhat the team already hada spreadsheet
My pipeline, and the spreadsheet the team already had.

The right size

As engineers, we tend to reach for the most impressive tool. I did, both times. But impressive is not the point. If a simple model does the job, and the people who use it can understand it, that is more than fine.

You do not need a spacecraft to commute between two small towns. And you do not need a container ship to send one small parcel.

A short trip between two small townsa spacecrafta carOne small parcela container shipa van
The trip decides the vehicle.

So now, before I choose a method, I ask how long the trip is, and how big the parcel is.

Let's talk.

If your question depends on real world health data, I probably want to hear about it.

Email me

TurkishMerhabaAzeri, my mother tongue, as written in Urmia: xoş gəldin, welcomeخوش گلدینAzerbaijani, as written in AzerbaijanSalamPersian, the language of school: dorudدرودEnglishHelloDutch, as said in FlandersDagyou
A greeting in each language I speak, where it lives.