Tuesday, October 2, 2018

Regression vs Classification in Machine Learning

REGRESSION :
  • The output variable takes continues values
  • This predicts a value from a continue set
  • Regression involves estimating or predicting a response
  • e.g. The price of a house depending on the 'size', 'location' can be some 'numerical value'. This numerical value can be continues. 
  • Given  f : x -> y,  if y is real number/continues, then this is a regression problem.
  • Regression models makes predictions that answer questions like the following:
    • what is the value of a house in Taipei, Taiwan ?
    • what is the probability that a user click on this ad ?
  • Different types of regression :
    • Linear regression
    • Logistic regression
    • Polynomial regression
    • Stepwise regression
    • Ridge regression
    • Lasso regression
    • ElasticNet regression
    • and many more ....

CLASSIFICATION :
  • The output variable takes class labels
  • classification predicts the "belonging" to the class
  • classification is identifying group membership
  • e.g. The prediction of house price can be classified into 'very costly', 'costly', 'affordable', 'cheap', 'very cheap'. Here each class may correspond to some range of values.
  • Given  f : x -> y,  if y is discrete/categorical variable, then this is classification problem.
  • Classification model predict discrete values that ans. the questions like the following:
    • Is the given email message spam or not spam ?
    • Is this an image of dog, a cat or a hamster ?
  • Different classification methods:
    • Linear classifier: Logistic regression, Naive Bayes classifier
    • Support Vector Machines
    • Decision Trees
    • Boosted Trees
    • Random forest
    • Neural Networks
    • Nearest neighbor
    • and many more ....



[src] :

No comments:

Post a Comment