Okay, so today I decided to mess around with this thing called “Mertens’ theorem.” I’d heard about it, but never really dug in, so I figured, why not? Let’s see if I can actually predict something with it!
Getting Started
First, I needed some data. I grabbed a dataset. It is a classfication dataset.
Then came the coding part. I’m not the best coder, to be honest, and I really just use python and some common libraries.
The Code Struggle
I loaded the data and started playing around. The first hurdle? Figuring out the core concept of Mertens. After finding its formula, I tried to translate formula into my simple code, so I define a function.
The Actual Prediction
After all the coding, I use my function.
I trained a base model like LogisticRegression, and I got my base model’s prediction, and I put the result into the mertens function. Finally I got a result from mertens function, it is a predicted value, looks good.
Did It Work?
Honestly, I was holding my breath. I compared my prediction to the actual results. And… it was pretty good!I was surprised that it works good!
So, that was my little adventure with Mertens’ prediction. I built it, I struggled with it, I tested it, and it actually worked (kind of!). It just goes to show, sometimes diving into something new can lead to some surprisingly cool results, even if you’re not an expert.
