7/ Challenges & Limitations
No Labels: Hard to evaluate model performance.
Interpretability: Discovered patterns may lack clear meaning.
Scalability: Large datasets require high computing power.
Algorithm Selection: The right method depends on the data.

6/ Real-World Applications
Customer Segmentation: Grouping users based on behaviour.
Fraud Detection: Spotting anomalies in transactions.
Topic Modeling: Extracting hidden themes from text.
Gene Expression Analysis: Finding patterns in biological data.

5/ Dimensionality Reduction Methods
PCA: Reduces features while keeping variance.
t-SNE: Maps high-dimensional data into 2D/3D.
Auto-encoders: Neural networks that learn compressed data representations.

4/ Key Clustering Algorithms
K-Means: Assigns data points to K clusters.
Hierarchical Clustering: Creates nested clusters.
DBSCAN: Detects clusters of varying density.
Gaussian Mixture Models (GMM): Assigns probability distributions to clusters.

3/ Types of Unsupervised Learning
Clustering: Finds natural groupings (e.g., customer segmentation).
Dimensionality Reduction: Simplifies data while retaining patterns (e.g., PCA, t-SNE).
Anomaly Detection: Identifies unusual data points (e.g., fraud detection).

2/ How It Works
The model processes input data (X) without labeled outcomes (Y). It detects similarities, differences, and structures. Two main tasks:
Clustering: Grouping similar data points.
Dimensionality Reduction: Compressing data while preserving key features.

7/ Challenges & Limitations
Data Dependency:Requires large, high-quality labeled datasets
Overfitting: Models may memorise training data instead of generalising
Bias n Variance: Finding the right balance is crucial
Scalability:Training large model demands high computational power

6/ Real-World Applications
Fraud detection in finance.
Medical diagnosis using patient data.
Customer churn prediction in business.
Sentiment analysis in NLP.
Image recognition for security and automation.

5/ Advanced Algorithms
Neural Networks - Inspired by the human brain, excels in complex pattern recognition.
Gradient Boosting (XGBoost, LightGBM) - Powerful for structured data and competitions.
K-Nearest Neighbours (KNN) - A simple but effective method for classification.

4/ Key Algorithms
Linear Regression - Models linear relationships between variables.
Logistic Regression - Used for binary classification.
Decision Trees - Splits data based on conditions.
Random Forests - An ensemble of decision trees.
SVM - Finds the best decision boundary.

3/ Types of Supervised Learning
Classification: Predicts categorical labels (e.g., spam detection, medical diagnosis).
Regression: Predicts continuous values (e.g., stock prices, temperature forecasting).

2/ How It Works
The model is trained on input-output pairs.
It identifies patterns and learns a function that generalizes to unseen data.
Performance is measured using metrics like accuracy, mean squared error, precision, and recall.