May 30, 20261 min read

An interactive guide to EDA

EDAData scienceTeaching

You open a new dataset and suddenly your brain goes: histogram? boxplot? scatter? heatmap? Same.

So instead of opening 17 tabs and re-watching the same videos, I built an interactive EDA guide.

Not another wall of text

No walls of text. Just visuals, interactions, and the flow that actually makes sense — from understanding your data to picking the right plot for the right question. It covers the complete EDA workflow, and yes, dark mode, because we're civilized.

The goal is simple: if it saves you from one more "wait… which plot do I use?" moment, it did its job.

Why I keep building these

Every one of these small interactive tools teaches me more than the tutorial it replaced. Explaining a thing visually forces you to actually understand it — there's no hand-waving when you have to draw it. That's the real reason they exist.

Keep reading

Making backprop visible

I understood backprop on paper, yet I couldn't see it happening. So I built a scroll-driven walkthrough where the network builds itself and the gradients flow backward in front of you.

Build it before you abstract it away

No PyTorch. No .fit(). Just NumPy and the chain rule — and why building a neural network by hand changed how I think about everything downstream.