Skip to content

01. Framing

Jump to bottom
Antonio Erdeljac edited this page Feb 26, 2019 · 1 revision

Framing

Topic: Framing

Course: GMLC

Date: 10 February 2019 

Professor: Not specified

Resources

Key Points

  • Supervised machine learning - A model trained to predict unlabelled data by combining labeled examples

  • Label - the y variable in linear regression (price, type, meaning etc.), the thing we are predicting

  • Feature - the x variable in linear regression (for a spam detector - words, address, time, phrases etc.), the thing we take in consideration when predicting

  • Example

    • instance of data, commonly held in data sets

    • 2 categories:

      • labeled  - {features, label} : (x, y) - what we train the model with

      • unlabelled - {features, ?} : (x, ?) - what the models guesses

  • Models  

    • Relationship between labels and features

    • Phases of a model’s life:

      • Training - teaching the model with labeled examples

      • Inference - applying the trained model to unlabelled examples

    • Types of a model:

      • Regression - predicts continuous values (value of something, probability of something)

      • Classification - predicts discrete values (spam or not spam, type of something, for example a cat, a dog, a cow etc.)

Check your understanding

  • Know the difference between labeled and unlabelled examples

  • Know the difference between a label and a feature

  • Identify phases of a model’s life

  • Identify types of a model

  • Know the definition of Supervised ML

Summary of Notes

  • Supervised ML is a model trained to predict never-before-seen examples by inference

  • Models can either be Regression type or Classification type

  • Label is the thing we are predicting

  • Feature is the thing we are taking in consideration when predicting

Clone this wiki locally