Supervised Learning: Classification and Regression Problems

In supervised learning, an information (called label) is attached to an object/observation (called Training data). The training data is consist of a set of training examples/labels.  Each example is a pair consisting of an input object and the desired output value.

Input object is typically a vector and the desired output is called as the supervisory signal.

Supervised Learning:

The process is to learn and predict for observations without labels.
– X, Y (Pre classified training examples)
– Given an observation X, what is the best label for Y?

Here, in this machine learning algorithm, we provide the information such as (X, Y), where every Y is a label for corresponding X.
With the help of different pairs of (X, Y) we create a model, after the training and tunning of the model, we provide new input to the model to find the best Yt for the input Xt.

It is called supervised learning because the process of an algorithm learning from the training dataset can be thought of as a Supervisor supervising the learning process.

Example:

Above image shows the example, though it can be further divided into Classification and Regression problems.

Classification Problems: In this type of Supervised Learning, the output is categorical. Such as,
1. True/False
2. Buy/Sell
3. Missing/Found
4. Dog/Cat
5. Orange/Banana.

Regression Problems: In this type of Supervised Learning, the output is a real value. Such as,
1. Height
2. Dollars
3. Weight
4. Temperature
5. Pointers
6. Salary

Some of the examples of Supervised Learning algorithms are:
• Decision trees
• Support vector machine (SVM)
• k-Nearest Neighbors
• Naive Bayes
• Random forest
• Linear regression
• polynomial regression
• SVM for regression

Summary:

Supervised Learning:

– X, Y (Pre classified training examples).
– Given an observation X, what is the best label for Y?

Classification: Assign a label.

Regression: Predict a continuous numerical value.