Machine Learning can be divided into three major types based on how the model learns from data.

Types of Machine Learning

Machine Learning
      │
      ├── Supervised Learning
      │
      ├── Unsupervised Learning
      │
      └── Reinforcement Learning

Types of Machine Learning

1. Supervised Learning

Definition

Supervised Learning is a type of Machine Learning in which the model learns using labeled data.

Labeled data means:

Input → Correct Output already exists

The model learns from examples and then predicts outputs for new data.

Working

Training Data
(Input + Output)
        ↓
Model learns relationship
        ↓
Prediction for new data

Example:

Student marks dataset:

Study HoursMarks235455670890

Model learns:

Study Hours → Marks

If:

Study Hours = 5

The model predicts:

Marks ≈ 62

Types of Supervised Learning

A) Regression

Predicts continuous values

Examples:

• House price prediction
• Temperature prediction
• Sales prediction

Algorithms:

• Linear Regression
• Polynomial Regression
• Ridge Regression
• Lasso Regression

B) Classification

Predicts categories/classes

Examples:

• Spam detection
• Disease prediction
• Email classification

Algorithms:

• Logistic Regression
• Decision Tree
• Random Forest
• Support Vector Machine (SVM)
• Neural Networks

Advantages

• High accuracy if labeled data exists
• Easy to evaluate

Disadvantages

• Requires large labeled datasets
• Labeling can be expensive

2. Unsupervised Learning

Definition

Unsupervised Learning is a type of Machine Learning where the model learns from unlabeled data.

Unlabeled data means:

Input exists
Output does not exist

The model tries to find hidden patterns and relationships.

Working

Input Data
(No output labels)
        ↓
Find hidden structure
        ↓
Groups or patterns

Example:

Customer shopping data:

CustomerAgeAmount SpentA22500B23550C504000D554500

The model may automatically create:

Group 1 → Young customers
Group 2 → Older high-spending customers

Types of Unsupervised Learning

A) Clustering

Groups similar data points.

Examples:

• Customer segmentation
• Image grouping
• Market analysis

Algorithms:

• K-Means
• Hierarchical Clustering
• DBSCAN

B) Association

Finds relationships between items.

Example:

People buying bread also buy butter

Algorithms:

• Apriori
• FP Growth

C) Dimensionality Reduction

Reduces features while preserving information.

Examples:

• Data compression
• Visualization

Algorithms:

• PCA
• t-SNE

Advantages

• No labeled data needed
• Finds hidden patterns

Disadvantages

• Harder to evaluate
• Results may be less interpretable

3. Reinforcement Learning

Definition

Reinforcement Learning is a type of Machine Learning where an agent learns by interacting with an environment and receiving rewards or penalties.

The model learns through trial and error.

Working

Agent
   ↓
Take Action
   ↓
Environment
   ↓
Reward/Penalty
   ↓
Learn

Example:

Teaching a dog:

Sit correctly → Reward
Wrong action → No reward

The dog gradually learns.

Another example:

Self-driving car:

Move correctly → +10 reward
Hit obstacle → -20 penalty

Components of Reinforcement Learning

1. Agent → learner
2. Environment → surroundings
3. Action → decision taken
4. Reward → feedback
5. State → current situation

Algorithms

• Q-Learning
• Deep Q Networks (DQN)
• SARSA
• Policy Gradient Methods

Advantages

• Learns automatically through interaction
• Useful for dynamic environments

Disadvantages

• Needs many training iterations
• Computationally expensive

Quick Comparison

SupervisedUnsupervisedReinforcementUses labeled dataUses unlabeled dataUses rewards and penaltiesPredicts outputsFinds hidden patternsLearns actionsTeacher availableNo teacherFeedback availableExample: Spam detectionExample: Customer segmentationExample: Self-driving car

Simple memory trick:

Supervised → Learn with answers

Unsupervised → Learn without answers

Reinforcement → Learn by rewards