DSML_Quiz

• Machine learning enables computers to learn patterns and make predictions from data without explicit programming.

Supervised vs Unsupervised Machine Learning

• Supervised learning involves training a model on a labeled dataset, where each input is paired with a correct output.
• The model learns to map inputs to outputs, allowing it to predict labels on new, unseen data.
• In unsupervised learning, a model is trained on an unlabeled dataset, where the algorithm learns to identify patterns, structures, and relationships in the data without explicit guidance.
• K-means and DBSCAN are types of clustering
• PCA or Principal Component Analysis is a type of Dimensionality reduction

Clustering

• Clustering is the process of grouping similar data points into clusters based on inherent similarities in the data.
• Clustering algorithms aim to maximize the similarity within clusters and minimize the similarity between clusters.

K-Means

• K-means clustering is an unsupervised learning algorithm that groups unlabeled data into distinct clusters based on feature similarity, with "K" representing the number of clusters.
• K-means identifies clusters by iteratively refining centroid positions, which are points representing the "centre" of each cluster.
• The algorithm assumes that data points closer to a centroid belong to the same group, mimicking how humans naturally categorize objects spatially.
• The goal of K-means is to minimize the within-cluster sum of squares (WCSS), a measure of the squared distances between data points and their respective centroid.
• The process involves initializing centroids, assigning points to clusters, and recalculating centroids until convergence is reached and clusters no longer move.
• Strengths of K-Means: simplicity, easy to implement and interpret, efficiency, has linear time complexity O(n·k.dt), and versatility
• Weaknesses of K-Means: manual selection, sensitivity to initialization, and assumption of spherical clusters.

K-Means Elbow Method

• It allows for optimal “k” selection (number of clusters
• K-Means aims to diminish the within-cluster sum of squares

DBSCAN

• DBSCAN or Density-Based Spatial Clustering of Applications with Noise identifies clusters as dense regions separated by sparse areas, excelling at detecting irregularly shaped clusters and automatically filtering noise.
• Key definitions for DBSCAN:
• Epsilon (ε)-neighborhood: All points within a specified radius ε of a given point.
• Core point: A point with at least a minimum number of (min_samples) neighbors within its ε-neighborhood.
• Border point: A non-core point that lies within the ε-neighborhood of a core point.
• Noise: Points that are neither core nor border points.
• Strengths of DBSCAN: Shape Flexibility, Noise Robustness, & Parameter Guidance
• Limitations of DBSCAN: Density Uniformity, High-Dimensionality, Border Point Ambiguity.
• DBSCAN handles non-convex shapes
• DBSCAN has explicit outlier detection
• Parameter Sensitivity is critical e/min_sample selection for DBSCAN.
• DBSCAN Struggles with varying densities
• DBSCAN Complexity: O(n log n) with spatial indexing

Dimensionality Reduction

• High-dimensional data often contains redundancies and noise.
• Dimensionality reduction simplifies such data.

Principal Component Analysis (PCA)

• PCA identifies latent variables as directions of maximum variance that encode the most informative features.
• Other dimensionality reduction methods include: Singular Value Decomposition (SVD), Non-Negative Matrix Factorization (NMF), Random projections, UMAP (Uniform Manifold Approximation and Projection), t-SNE (t-Distributed Stochastic Neighbor Embedding), Independent Component Analysis (ICA), and Feature Agglomeration