• Contents:

ML - Supervised Learning

Ref: https://www.udemy.com/course/aws-ai-practitioner-certified/learn/lecture/44886601

• 🔧 Learn a mapping function
  • Can predict output for new unseen input data
• Needs labeled data
  • very powerful
  • difficult to perform on millions of datapoints

Regression Algorithms

• 🔧 Predicts numeric value based on input data
  • Continuous output variable
• 💡Linear regression (straight line to describe relationship) very common, but many other types of regression
• Examples
  • Predicting House Prices
  • Stock Price Prediction
  • Weather Forecasting
• Simple linear regression diagram

Classification Algorithms

• 🔧 Predicts categorical label of input data
  • Discrete output variable
• Types and examples
  • Binary Classification (e.g. classify card transactions as “fraud” or “not fraud”)
  • Multi-class Classification (e.g. classify fruits as "orange", "apple", or "pear”)
  • Multi-label Classification (e.g. assign multiple labels to a movie, like "action" and "comedy”)
• Example of a classification algorithm: kNN (k-Nearest Neighbors) → give a datapoint the label(s) of the k closest datapoints
  • 💡 kNN can also do regression → give a datapoint the average values of the k closest datapoints

Training, Validation, and Test Sets

1. Training Set: trains the model
2. Validation Set: validates model performance between training iterations
   • ❗ Validation sets are optional
3. Test Set: evaluates final model performance (after all training has been done)

• 💡 Typical percentages of original dataset:
  • 60-80% Training
  • 10-20% Validation
  • 10-20% Test