What is Supervised Learning in Machine Learning and its types.

Supervised Learning

Supervised machine learning is a subcategory of machine learning and artificial intelligence.

It’s one of the introductory algorithms in machine learning concepts and is most commonly used for data analysis.

Supervised Learning algorithms have decisive output. Comparing models is easier in supervised Learning and is done using prediction Error.

Supervised Learning is done in two ways, Regression and Classification.

Regression

Regression is used when the output variable is continuous in nature or a real value. Examples are predicting a stock, pricing house/car.

Types of Regression:

Simple Linear Regression

This regression is used for understanding the relationship between two variables. One variable is independent and the other is a dependent variable.

Formula : y= mx+c

Polynomial Regression

This regression is used for predicting a value using 2 or more independent variables.

Y = a+b1x1+b2x2+…..

Ridge Regression

It is an extension of linear regression that adds a regularization penalty to the loss function during training.

It is a technique of analyzing multiple regression data that suffer from multicollinearity. Multicollinearity is the occurrence of high correlations among two or more independent variables.

To reduce multicollinearity and high variance, squared bias is added to the loss function.

Uses L1 regularisation technique.

Lasso Regression

Lasso Regression also adds bias to reduce the effect of multicollinearity and high variance like Ridge Regression, but here, absolute value is added instead of squared bias.

Lasso regression not only helps in reducing overfitting but it can help us in feature selection.

Uses L2 regularisation technique.

Used for accurate prediction

Lasso regression uses shrinkage, where the data values are shrunk towards a central point such as the mean value.

Classification

Classification is used when the output variable is categorical. Examples are spam/not spam, male/female.

Types of classification:

Binary classification

Binary Classification is the simplest of all the machine learning algorithms.

The given data is classified into two classes. eg: Male or Female. True or False.

Logistic Regression is the most commonly used algorithm for binary classification.

Logistic Regression

A logistic regression model predicts a dependent data variable by analyzing the relationship between one or more existing independent variables.

The logistic function is also called the sigmoid function.

Sigmoid function e — Euler’s number

Multi Label classification

This is used when there are three or more classes and the data belongs to one or more class labels. Example : photo classification. In a photo, there can be multiple objects. So each output will have multiple objects.

Popular algorithms that are used for Multi Label Classification are :

1. Multi-Label decision trees
2. Multi-Label RandomForest
3. Multi-Label Gradient Boosting

Multi Class Classification

This is used when there are three or more classes for the output label. And the data belongs to only those classes. Example : Face Classification.

Popular algorithms that are used for Multi Class Classification are :

1. Decision Tress
2. KNN
3. Random Forest
4. Gradient Boosting