Unsupervised learning

In summary, unsupervised learning is about discovering hidden patterns in data without predefined labels, making it valuable for exploratory data analysis, clustering, and Anomaly Detection.

Unsupervised learning is a type of machine learning where the algorithm is trained on data without explicit labels or predefined outputs. The goal of unsupervised learning is to find hidden patterns, relationships, or structures in the data. Unlike supervised learning, where the model learns from labeled input-output pairs, unsupervised learning only uses input data, and the algorithm attempts to uncover insights from the data by itself.

Key Concepts of Unsupervised Learning:

1. No Labeled Data: There is no ground truth or correct output associated with the input data.
2. Data Patterns: The algorithm identifies inherent structures, clusters, or associations within the dataset.
3. Objective: The primary objective is to explore the data and organize it in a way that makes the underlying patterns more apparent.

Common Types of Unsupervised Learning:

1. Clustering:

   • The algorithm groups similar data points together based on their features.
   • Example: Customer segmentation in marketing. A clustering algorithm can divide customers into different groups based on purchasing behavior, demographics, or browsing history.
   • Popular Algorithms:
     • K-means Divides the data into $k$ clusters, where each data point belongs to the nearest cluster.
     • Hierarchical clustering: Builds a tree of clusters by progressively merging or splitting groups based on similarity.
     • DBSCAN (Density-Based Spatial Clustering of Applications with Noise): Forms clusters based on data density, allowing it to find arbitrarily shaped clusters.

   Example: Clustering customers into distinct groups based on their purchasing patterns in an e-commerce business.

2. Dimensionality Reduction:

   • Reduces the number of input variables (features) while preserving as much information as possible. This is helpful when dealing with high-dimensional data, where visualization and analysis become challenging.
   • Popular Algorithms:
     • Principal Component Analysis (PCA): Transforms the data into a set of orthogonal components that explain the maximum variance. Example: PCA used to reduce the dimensions of a dataset containing hundreds of features, such as a dataset of pixel values from images.

3. Anomaly Detection

   • Identifies outliers or unusual data points that don’t conform to the expected pattern in the dataset.
   • Example: Detecting fraudulent credit card transactions by identifying transactions that deviate significantly from typical spending patterns.
   • Popular Algorithms:
     • [[Isolation Forest]]: Works by randomly partitioning the data and identifying anomalies as points that can be isolated quickly.

What is Supervised Learning?;; Supervised learning is a process where learning algorithms are trained on labeled datasets, where each point is associated with a class or label. That is there is a y_train, then uses to get y_pred (from X_test) and compare against y_test.

A Supervised task that involves ;; assigning predefined categories to an input. The goal is to learn a mapping via training so one can predict categories/labels of unseen data.

Has labelled data.

Examples:

Support Vector Machines K-nearest neighbours

Unsupervised Learning

: Unsupervised learning algorithms learn from unlabelled data, where the goal is to discover the underlying structure of the data. These algorithms can be used for clustering and dimensionality reduction.