Model Anatomy · Episode 07 · 1:23
Linear regression from scratch in Python, no libraries
Start with a flat line through 10 delivery trips, measure each error and nudge w and b for 1,000 rounds. It learns minutes = 2.98 × km + 4.21 and predicts a 12 km trip at 39.9 minutes.
What you’ll needPython 3.10+
What you’ll need
- python.org/downloads →
Python 3.10 or newerRuns the code. Check yours with python3 --version. - code.visualstudio.com →
VS Code or CursorEither works; so does any editor you like. - git-scm.com/downloads →
gitGets the code from GitHub. - No packages needed: it uses only the Python standard library.
Set it up
- 1Get the code
git clone https://github.com/DayanEbrar0X/data-anatomy.ai.git cd data-anatomy.ai
- 2Make a virtual environmentIt keeps this project's packages separate from the rest of your computer.
python3 -m venv .venv source .venv/bin/activate
- 3Run it
cd machine-learning/07-linear-regression-no-libraries python3 src/delivery.py
Project structure07-linear-regression-no-libraries/
Project structure
machine-learning/07-linear-regression-no-libraries/Open on GitHub →
- short/
- learn.pythe 9-line version from the short
- src/
- delivery.pymainthe 17 lines from the video
- README.mdthe idea, how to run it, a line-by-line walkthrough, things to try
machine-learning/07-linear-regression-no-libraries/src/delivery.pyOpen on GitHub →
km = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
mins = [8, 9, 15, 15, 20, 20, 26, 27, 31, 35]
w, b = 0.0, 0.0
lr = 0.02
n = len(km)
for epoch in range(1000):
dw, db = 0.0, 0.0
for x, y in zip(km, mins):
err = w * x + b - y
dw += err * x / n
db += err / n
w -= lr * dw
b -= lr * db
print(f"mins = {w:.2f} * km + {b:.2f}")
print(f"12 km: {w * 12 + b:.1f} min")Transcript
The loop that trains every neural network fits in fifteen lines. No libraries. Guess, check, nudge, repeat. Say we're predicting delivery times.
Ten past trips: kilometers and minutes. The model is just a line: w tilts it, b lifts it. Both start at zero: a flat line. lr sets the nudge size.
n counts the trips. Now, a thousand rounds of practice. Each round opens two tallies: tilt, and lift. Then we visit every trip.
The line guesses: w times x, plus b. Minus the real time: the error. Eight minutes short. Each error votes on the tallies, averaged over ten trips.
Then w and b take a small step against the error. Like a shower: too cold? Turn it a bit. Check again.
Round after round, its average miss drops to about one minute. Print the line and one prediction. Let's run it. It learned: 2.
98 minutes per kilometer, plus 4.21. A new 12 km order? 39.
9 minutes. The same loop trains every neural network. Just billions of numbers, not two. Plain lines still price and forecast: fast, and explainable.
One catch: set lr too big, and it overshoots. Guess, check, nudge. Fifteen lines, and it learns.
Read the lesson on GitHub →- #linearregression
- #machinelearning
- #python
- #datascience
