Supervised Unsupervised Learning

Supervised Learning

• "Right answers" given

Types of Supervised Learning

Regression Problem

• Predict continuous valued output

  • e.g. can be between 200 and 300 dollars

• Plotting Data: Think of line of best fit on graph to predict

  • e.g. House prices (1 feature to predict price - square footage)

    • X-axis: Square footage

    • Y-axis: Price

    • Probably a positive correlation (can be fitted with line, curve, etc.

Classification Problem

• Discrete value output

• e.g. Breast cancer classification - In relation to tumor size (1 feature)

  • Can be 0, 1, 2, or 3 - malignant, type 1, type 2, etc.

Unsupervised Learning

• "No correct anwers given"

• Derive structure from data

• No feedback based on prediction results

Types of Unsupervised Learning

Clustering Algorithm

• Given a dataset, can you find structure in the set?

• Group together elements related with different variables

  • e.g. lifespan, location

Non-Clustering Algorithm

• Aka Cocktail Party Problem Algorithm

• Find structure in chaotic environment

  • e.g. Two voices in one audio recording - We can separate them by identifying the different sounds