The Machine Learning Starter Kit — from your first line of Python to deployed models
A 172-page machine learning starter kit with 26 chapters. Every code listing was actually executed — the outputs on the page are the real ones.
Get the Starter Kit — €19 → Instant download · PDF · Yours forever
Code that ran
Every listing was executed before printing. What you see is what Python returned.
26 chapters
From Python basics through algorithms, evaluation, and deployment — in one arc.
57-term glossary
Every ML term you’ll hit in docs and interviews, defined in plain language.
We print the real output. Even when it surprises us.
Most ML books show idealized results. In the Starter Kit, every listing was executed and its output captured verbatim — including the chapter where plain logistic regression beat both tree ensembles.
That’s not a bug. Learning why it happened is the lesson.
# Listing 16.1 — boosting, alongside its rivals gb = GradientBoostingClassifier( n_estimators=200, learning_rate=0.05, max_depth=3, random_state=0 ) gb.fit(X_train, y_train) gb.score(X_test, y_test) logistic regression test acc=0.870 random forest test acc=0.830 gradient boosting test acc=0.840 # Yes — the simplest model won. Chapter 16 # explains exactly why.
A real excerpt: The Sacred Test Set
Selected passages and Listing 12.1 from Chapter 12, printed pages 82–83 of the Machine Learning Starter Kit. Reformatted for reading here; this is an excerpt, not the complete chapter. No email required.
That box is your test set. Everything you do for the rest of the project, cleaning decisions, feature engineering, choosing a model, tuning it, happens using only what remains, the training set. The test set is opened exactly once, to report a single, honest number for how the finished model performs on data it has never influenced in any way.
from sklearn.model_selection import train_test_split
# Do this FIRST, before exploring, before tuning, before anything
X_train, X_test, y_train, y_test = train_test_split(
X, y, test_size=0.25, random_state=0
)
# 🔒 X_test and y_test are now sealed.
# Every decision from here on uses ONLY X_train and y_train.Why this is sacred
If you use the test set to make even one decision, which features to keep, which model to prefer, which setting performs best, you have leaked information into your process, and the number it eventually reports stops being honest. It is the difference between a real exam and one where you have already seen the answer key.
Reading note: this random-split illustration is for suitable independent observations. Time-ordered or grouped data needs an appropriate split. Use validation data or cross-validation within the training data to choose models; reserve the final test set for final evaluation.
Continue with the complete handbook: 26 chapters covering Python, data preparation, modelling, evaluation, and deployment.
Get the complete Starter Kit — €19 →What’s inside the Machine Learning Starter Kit
26 chapters · 6 parts · one continuous build from zero to deployed
The Python Toolkit
Your environment, Python essentials, NumPy and pandas, messy real data, and moving from notebooks to software.
The Mathematics
Describing data, distributions and skew, probability and Bayes, vectors and matrices.
How Models Learn
What a model is, supervised vs unsupervised, the sacred test set, and evaluating honestly.
Building Real Models
Decision trees, random forests, gradient boosting, and feature engineering.
Deep Learning & Specialisation
Your first neural network, text as numbers, images as arrays, and choosing where to specialise.
Shipping
Packaging a model, serving predictions with FastAPI, monitoring in production, and what comes next.
Want to go deeper on any topic? The Kit pairs well with our curated machine learning books, ML online courses, free datasets for practice, and ready-to-use code snippets.
Skip the tutorial maze. Start executing.
One PDF, 172 pages, every line of code tested. Buy once, keep forever, learn at your own pace.
Get the ML Starter Kit → Secure checkout via Gumroad · Instant PDF deliveryWho is the Machine Learning Starter Kit for?
The Kit is written for people who learn by running code, not just reading about it. It’s a perfect fit if you are:
- A roadmap follower: you’re working through the free 90-Day Roadmap and want the code companion.
- A hands-on learner: you understand things when you can run them, break them, and fix them.
- A career changer: you need working projects, not just theory, to show employers.
- A busy self-learner: you’d rather pay €19 than spend weeks assembling scattered tutorials.
Frequently Asked Questions
How is this different from the free 90-Day Roadmap?
The roadmap is a complete, free strategy: it tells you what to learn each day and links you to resources. The Starter Kit is the execution shortcut: 172 pages of explained, tested code so you don’t have to assemble it yourself. You can absolutely succeed with just the roadmap — the Kit saves you time.
What format is the book?
A downloadable PDF, delivered instantly after purchase through Gumroad. Read it on any device; it’s yours forever, including future updates to this edition.
Do I need prior experience?
No. Chapter 0 starts by setting up your Python environment from scratch. If you already code, the early chapters go quickly and the algorithm and evaluation chapters are where you’ll spend your time. If you’d like extra Python practice alongside the book, check our Python courses and Python books pages.
Is the code really executed?
Yes — every listing was run before printing, and the outputs on the page are captured verbatim from those runs. When a result was surprising, we kept it and explained it rather than swapping in a prettier number.
What tools do I need?
Any laptop that can run Python. Everything in the book uses free, open-source tools: Python, uv, JupyterLab, pandas, scikit-learn and FastAPI. No paid software.
What comes after the Starter Kit?
Once you’ve finished the book, keep building: work through our free tutorials, practice on real datasets, and specialize with our curated statistics courses and AI books.
Can I get a refund?
If the Kit isn’t what you expected, reply to your purchase email within 14 days and we’ll refund you. No forms, no friction.
Not ready to buy? Start with the free Roadmap.
The 90-Day ML Roadmap is completely free and genuinely complete. Get it, start learning, and come back for the Kit if you want the code companion.