Getting Started with Machine Learning Basics

Starting with machine learning can feel overwhelming — where do you begin, what do you need, and how do you stay motivated? This post shares a clear, student-friendly roadmap inspired by Jason Brownlee’s famous "Machine Learning Mastery" guide, along with my own thoughts as a learner. Let's break it down together.

1. Adjust the Mindset

The most important thing and the first thing to do when you try to learn ML or any Other Technologie or anything is to Avoid Self-Limiting Beliefs, "What if I fail? Or I'm not good enough?..." . You have a brain, so you are capable of learning anything. , Our brains are like muscles—they adapt and grow with practice. Don’t stop yourself because of requirements—you don’t need a fancy CPU/GPU to start or to be a good programmer. Just start.

No more excuses! And let's dive into the main Parts of How to get start with machine learning Using Jason Brownlee's approach .

2. Find Your Tribe

Tribe is the group you are belong two and there are 10 mains groups of people interested in ML.

I’ll give you a quick overview about each Tribe

Business Tribes

People in this group are focused on using machine learning (ML) in their companies, without needing to know the deep technical stuff.

Business Person (General Interest)
- Execs or consultants curious about ML.
- Want to use ML in strategies or future projects.
- Helpful resources: Gartner reports, McKinsey guides, and beginner-friendly books for managers.
Project Manager (Delivering a Project)
- Leaders overseeing ML projects.
- Want big-picture knowledge, not technical details.
- Recommended books: Predictive Analytics, Data Science for Business, Data Smart.

Academic Tribes

These are students, professors, or researchers deeply interested in the science behind ML.

Student (Undergrad/Grad)
- Learning ML in class, diving into specific algorithms.
- Best with structured learning and textbooks like:
  - Machine Learning: A Probabilistic Perspective
  - Pattern Recognition and Machine Learning
Researcher (Advancing the Field)
- Focused on contributing new knowledge.
- Reads lots of research papers and attends top ML conferences (like NIPS, ICML).
- Prefers journals to textbooks.
Researcher (Solving a Specific Problem)
- Not focused on ML itself but wants to apply it (e.g., biologists, geologists).
- Values simple and explainable models (like linear regression).
- Needs a solid, systematic approach.

Engineering Tribes

These are developers and engineers looking to build smart software with ML.

Programmer (Learning Algorithms)
- Wants to understand ML by coding algorithms from scratch.
- Suggested books: Data Science from scratch, Machine Learning in Action.
Developer (Making One-Off Models)
- Not an ML expert, but wants to build a model to solve a business problem.
- Recommended resources: Applied Predictive Modeling, Data Mining: Practical ML Tools.
Engineer (Building Smart Services)
- Aims to add ML into production-ready apps.
- Balances performance, speed, and reliability.
- Uses libraries like Scikit-learn, and books like:
  - Learning scikit-learn
  - Practical Data Science with R

Data Tribes

These people focus on data first but use ML to get better insights.

Data Scientist (Solving Business Problems)
- Already uses ML, needs to stay up-to-date.
- Uses both applied and theoretical resources.
- Good reads: ISLR, Applied Predictive Modeling.
Data Analyst (Explaining Data Clearly)

Mainly explains data, occasionally uses ML.
Prefers simple, clear models (like regression).
Focus is on explainability over model complexity.

Still not sure where you belong?
Check out the full guide 👉 Find Your Machine Learning Tribe or ask ChatGPT/Gemini — there are no strict rules! 😊

Confused GIF

3. Pick a Tool

Select a Tool Based on Your Skill Level and Map It to Your Workflow

Beginner – Weka Workbench
Weka is a user-friendly, GUI-based machine learning tool ideal for beginners. It allows you to load datasets, preprocess data, build classifiers, and evaluate results without writing code. Perfect for quick experiments with supervised learning models such as decision trees, Naive Bayes, or SVM.
Intermediate – Python Ecosystem (Scikit-learn, TensorFlow, Keras, Pandas)
Python offers a flexible and scalable environment with powerful libraries for data processing, feature engineering, model training, and deployment. If you’re comfortable writing code, the Python ecosystem enables full control over your ML pipeline and is widely used in production systems.
Advanced – R Platform
The R programming language is popular among statisticians and advanced researchers. It offers robust libraries for data mining, statistical modeling, and machine learning. R is particularly useful when you need in-depth statistical analysis or want to integrate ML with advanced visualization and reporting.

📝 Note from a student:
I'm not a professor or a data scientist (yet!), but as a student, I would personally recommend starting directly with Python — or at least learning Python early — and then using it as your main tool instead of Weka. Of course, everyone learns differently. Some people prefer to take baby steps and start with a visual tool like Weka, and that’s totally okay. This is just a friendly recommendation from a fellow learner — feel free to follow the path that suits you best!

4. Practice on Datasets

The best way to learn machine learning is by applying it to real data. Here's how to get started:

1. Start with Small Datasets
Use simple, well-known datasets like the Iris dataset, Titanic survival, or handwritten digits . These are easy to load and train in memory, perfect for beginners.

2. Explore Real-World Problems
Try datasets from actual domains like health, finance, or cybersecurity. This helps you understand how messy real data can be — and how machine learning solves real challenges.

3. Pick Topics You Care About
If you enjoy what you're working on (e.g. sports stats, movie reviews...), you’ll stay motivated and learn faster.

Tip: You can find many free datasets on platforms like Kaggle, UCI Machine Learning Repository, or scikit-learn’s built-in datasets.

5. Build a Portfolio

Complete small, focused projects and demonstrate your skills .

The benefit of ML Portfolio :

For you:
- each project has a well-defined purpose
- completed project provide knowledge
- looking for consistent collection of projects can be used as a lever to keep you on Trade
To others:
- A project demonstrate the capability with regard to a specific problem domain,, tool, library or algorithm
- project must be understood at least in terms of its purpose and findings→ Understand the project that you put so that You can demonstrate your ability to communicate technical subjects well.
To community
- Public project means feedbacks , people seeing what you did , what you are trying to build , and how you build it.
- public project provide a point of study, perhaps for a specific algorithm or a problem

what is a good ML portfolio?

The portfolio is you , not just a collection of projects to show, it has to represent you and represent what you know what you can do.

Jason Brownlee set 5 properties for an effective portfolio:

Accessible : Make the portfolio public , people can find ,read ,comment on & use your work (!if possible)
small Projects: Each Project should be small in terms of effort ,resources and time
completed : Small project helps finish projects . set an objective and achieve it .
independent : Each project should be independent so it can be explained and understood in isolation
Understandable : Project Must clearly and effectively communicate its purpose and findings

Conclusion

Getting started with machine learning isn’t about having the perfect computer or background—it’s about mindset, community, and consistent practice. Identify your tribe, pick the right tools, get your hands dirty with real data, and build a meaningful portfolio that reflects your journey. There’s no single path to becoming a machine learning expert, but if you follow these steps and stay curious, you’ll get there.

Remember: Start simple. Start now. Learn loud. 🚀

If this guide helped you, and you'd like to support my learning journey or say thanks, feel free to buy me a coffee ☕. Your support means a lot! 💙

Machine Learning (Part 3) How Do I Get Started, inspired by My Take on Jason Brownlee