Machine Learning

Machine Learning Techniques

• Supervised ML
• Unsupervised ML
• Reinforcement Learning

Supervised ML Technique

• We have dataset with dependent(output) and independent variables(input).
• We train the model using data and then make predictions using trained model.
• E.g. we have a dataset with No. of rooms(independent variable) in house vs house price(dependent variable).
• Theere are 2 types of problem statement in Supervised ML technique:
  1. Regression problem statement
     • The dependent variable is continous.
     • E.g. the house price example above.
  2. Classification problem statement
     • The dependent variable is categorical.
     • E.g a data with no. of study hours(input) and pass or fail(output).
     • In this example we have only two category in output(pass or fail). Thus, called Binary Classification
     • If ther is multiple category like(pass or fail or maybe) then it is calles MultiClass Classification
• ALGORITHMS:
  • Linear Regression (Regression)
  • Ridge & Lasso (Regression)
  • ElasticNet (Regression)
  • Logistic Regression (Classification)
  • Decision Tree (Both Regression and Classification)
  • Random Forest (Both Regression and Classification)
  • AdaBoost (Both Regression and Classification)
  • Xgboost (Both Regression and Classification)

Unsupervised ML Technique

• We don't know the output feature.
• We don't need to predict anything. Instead, we need to find out similar clusters or groups.
• E.g. Customer Segmentation - we have a data of salary of people and their spending score, then we create differnt clusters of people like high salary low spending people, high salary high spending people etc. And e-commerce website have to send a email with discount coupon it will share it with a specific cluster of people.
• ALGORITHMS:
  • K Means
  • Hierarichal Mean
  • DB Scan clusteirng

Reinforcement Learning

• Learns itself.