ML - Field of study that gives computers the ability to learn without being explicitly programmed.

ML Algorithm

Supervised Learning

• Regression model - Predict real-valued output or continuous variables
• Classification model - Predict discrete valued output or predefined class

Unsupervised Learning

• Cluster
• Grouping

Reinforcement Learning

Model - Mathematical representation of a real-world process to establish relationship between features and label. ML model is generated by applying training data to a ML algorithm to learn from.

• Training - creating or learning the model
• Inference - applying the trained model to unlabeled examples to make useful predictions

Linear Regression with one variable

• Hypothesis h(x) = θ₀ + θ₁x where θᵢ is parameter, h is function of x for fixed θᵢ
• Cost Function J(θ₀,θ₁)=1/2m Σ(h(x) - y)² where m is number of training examples, J is function of θᵢ
• x > Input variable - Feature - individual independent variables that act as the input for prediction
• y > Output variable - Label - Prediction - example - future price of wheat
• Gradient descent - algorithm to find minimum of cost function J(θ₀,θ₁)

labeled examples: {features, label}: (x, y)

unlabeled examples: {features, ?}: (x, ?)

Tools

• Azure ML Studio - Web-based IDE that simplifies development and deployment of machine learning models and services.

Tutorial

• https://developers.google.com/machine-learning/crash-course

Jupyter Notebook online

• https://notebooks.azure.com
• https://colab.research.google.com