Modules, Packages, Virtual Environments and Git

import math                                  # whole module, use with prefix
from statistics import mean, median          # specific names
import datetime as dt                        # alias
from collections import Counter as C
print(math.sqrt(144), math.pi, mean([3, 5, 10]), median([3, 5, 10]), dt.date(2025, 4, 1).strftime("%d %b %Y"), C("aab"))
print(math.ceil(4.2), math.floor(4.8), math.log10(1000), round(math.exp(1), 4), math.comb(5, 2))

from pathlib import Path
Path("metrics.py").write_text('''"""Reusable business metrics."""
GST = 0.18

def margin_pct(profit, sales):
    return round(100 * profit / sales, 1) if sales else None

def with_gst(amount):
    return round(amount * (1 + GST), 2)

if __name__ == "__main__":          # runs only when executed directly, not when imported
    print("self-test:", margin_pct(25, 200))
''', encoding="utf-8")
import importlib, sys
if "" not in sys.path: sys.path.insert(0, "")        # notebooks and scripts already search the current folder
import metrics
importlib.reload(metrics)
print(metrics.margin_pct(30, 120), metrics.with_gst(1000), metrics.__doc__)
import subprocess, sys
print(subprocess.run([sys.executable, "metrics.py"], capture_output=True, text=True).stdout.strip())
Path("metrics.py").unlink()

churn-project/
├── pyproject.toml          # project metadata and dependencies
├── README.md
├── .gitignore              # venv, data, secrets, __pycache__
├── data/                   # raw/ and processed/ (usually not committed)
├── notebooks/              # exploration
├── src/churn/
│   ├── __init__.py
│   ├── load.py             # read and validate data
│   ├── features.py         # feature engineering
│   └── train.py            # model training
└── tests/
    └── test_features.py

import os, random, statistics, datetime as dt, itertools, textwrap, uuid, time
random.seed(42)
print(random.randint(1, 6), random.choice(["N", "S", "E"]), random.sample(range(100), 3), round(random.random(), 3))
print(statistics.stdev([2, 4, 4, 4, 5, 5, 7, 9]), statistics.mode([1, 2, 2, 3]))
today = dt.date(2025, 10, 27); due = today + dt.timedelta(days=45)
print(due, due.strftime("%A"), (dt.date(2026, 1, 1) - today).days, "days to new year")
print(os.cpu_count() > 0, os.getenv("HOME") is not None, textwrap.shorten("A very long product description here", 20), len(str(uuid.uuid4())))

# Create and activate an environment (built-in venv)
python -m venv .venv
source .venv/bin/activate          # macOS/Linux
.venv\Scripts\activate             # Windows

# Install, list, pin, reproduce
pip install pandas scikit-learn matplotlib
pip list
pip freeze > requirements.txt
pip install -r requirements.txt

# uv: a much faster modern alternative (by Astral)
uv init churn-project && cd churn-project
uv add pandas scikit-learn
uv run python src/churn/train.py

git init                              # start tracking a folder
git status                            # what changed?
git add analysis.py                   # stage a file  (git add . = everything)
git commit -m "Add churn EDA"         # save a snapshot with a message
git log --oneline                     # history
git switch -c feature/rfm             # new branch
git push -u origin feature/rfm        # upload the branch to GitHub
git pull                              # download others' changes
git clone https://github.com/user/repo.git
git diff                              # unstaged changes
git restore file.py                   # discard local changes to a file

.venv/
__pycache__/
.ipynb_checkpoints/
data/raw/
*.env
.DS_Store