Apr 13, 2026 · Day 2 of 30
Day 2: I met NumPy and Pandas today here's what actually made sense
Yesterday I committed to learning AI/ML in public. Today I did the first real thing I opened Python and started looking at the two libraries everyone mentions when talking about ML: NumPy and Pandas.
No model training yet. No fancy algorithms. Just getting comfortable with the tools that sit underneath all of it.
Why these two first?
From everything I've read, you can't really do ML without understanding how data is stored and manipulated. NumPy and Pandas are how that happens in Python. So before I touch any ML library, I want these to feel natural.
NumPy
Works with numbers at speed. Think of it as Python lists but way faster and built for math operations on large datasets.
Pandas
Works with tables of data. Like Excel but in Python rows, columns, filters, and a lot more control.
What I actually tried
I started with NumPy arrays creating them, doing basic operations, understanding why they exist when Python already has lists. The speed and simplicity of doing math on an entire array at once was the thing that clicked for me:
import numpy as np arr = np.array([10, 20, 30, 40, 50]) print(arr * 2) # Output: [20 40 60 80 100]
With regular Python you'd need a loop for that. With NumPy it's one line. That's when I understood why ML uses it you're constantly doing operations on thousands of numbers at once.
Then I moved to Pandas. I created a small DataFrame just to see how it feels:
import pandas as pd data = {"name": ["Alice", "Bob", "Kanishka"], "score": [88, 92, 79]} df = pd.DataFrame(data) print(df)
The moment the table printed cleanly in my terminal rows, columns, index — something about it felt satisfying. This is what real data will look like.
What still confuses me
Pandas has a lot of methods. A lot. I kept seeing things like .iloc, .loc, .groupby and I didn't go deep on any of them yet. That's tomorrow's problem. Today I just wanted to get comfortable with the shape of things.
One thing I'd tell Day 1 me
Don't try to memorise the methods. Just run the code, see the output, understand what changed. The rest comes with repetition.
QUESTION FOR YOU
Is there a Pandas or NumPy trick you wish someone had shown you early on? I'm at the very beginning genuinely curious what matters most.
Top comments (0)