Machine Learning Basics

Explore & Discover

Machine learning is already running your life — it picks your YouTube recommendations, filters your spam, and unlocks your phone with your face. Start by exploring how it actually works without any code. Check out Google's free "Teachable Machine" at teachablemachine.withgoogle.com — you can train an image classifier in your browser in under five minutes using your webcam. Try teaching it to recognize two different hand gestures. Also watch 3Blue1Brown's "But what is a neural network?" on YouTube — it's the clearest visual explanation out there. You're ready for the next step when you can train a basic Teachable Machine model and explain in your own words what "training data" means.

Learn the Basics

Now learn the core vocabulary. Head to fast.ai's free "Practical Deep Learning for Coders" (the first two lessons are accessible to beginners) and Khan Academy's AI and machine learning unit. Key terms to understand: model, dataset, training, testing, overfitting, accuracy, and labels. Learn the difference between supervised learning (you give it labeled examples) and unsupervised learning (it finds patterns on its own). No coding required yet — just build your mental map. You're ready for the next step when you can define training data, model, and accuracy in your own words and give a real-world example of supervised learning.

Build Your First Project

Write your first ML code using Python in Google Colab (free, runs in your browser — no installation needed). Find the "Scikit-learn Iris flower dataset" tutorial — it's the classic first ML project. You'll load a dataset of flower measurements, split it into training and testing sets, train a simple classifier, and check how accurate it is. Copy the code, run it, and change one variable at a time to see what breaks. freeCodeCamp has a free Python intro if you need a syntax refresher. You're ready for the next step when you can run the Iris classifier notebook in Google Colab and read the accuracy score it produces.

Experiment & Iterate

Pick a different dataset and train your own model from scratch. Kaggle.com offers hundreds of free beginner datasets — try something fun like predicting movie ratings or classifying types of music. Experiment with changing the algorithm (try k-nearest neighbors vs. a decision tree), adjusting how much data you use for training vs. testing, and see how each change affects accuracy. Keep a simple experiment log: write down what you changed and what happened. You're ready for the next step when you have trained a model on a dataset you chose yourself and can explain why your accuracy score went up or down after one change you made.

Advanced Techniques

Go deeper with neural networks. Use TensorFlow Playground (playground.tensorflow.org) — a free interactive visualizer that shows you how neurons connect and learn in real time. Then try building a simple image classifier using TensorFlow and Keras in Google Colab, following fast.ai Lesson 1. Learn about epochs, loss functions, and activation functions. Read about how Utah tech companies like Podium or Pluralsight use ML in their products — machine learning isn't just Silicon Valley. You're ready for the next step when you can run a TensorFlow image classifier, interpret the training loss graph, and explain what an epoch is.

Final Project Showcase

Build an original ML project that solves a problem you actually care about. Ideas: a model that classifies Utah wildflowers from photos, predicts snowfall at a Wasatch ski resort from historical data, or identifies spam vs. real comments. Document your process: what problem you chose, what data you used, how you trained your model, and what your accuracy was. Present your project in a Google Slides deck or a short recorded video walkthrough. Share your Colab notebook link so others can run it. You're ready for the next step when you have a working ML project with documented results that someone else can reproduce by running your notebook.