More Related Content
Similar to DB_ALGOS.pptx IT IS THE PPT CLUSTERING ; (20)
Recently uploaded (20)
DB_ALGOS.pptx IT IS THE PPT CLUSTERING ;
- 2. DBSCAN (Density-Based Spatial Clustering of Applications with Noise) • Concept: DBSCAN groups points that are closely packed together (points with many nearby neighbors), marking points in low-density regions as outliers. • Key Parameters: – ε (epsilon): Defines the radius of a neighborhood around a point. – MinPts: Minimum number of points required to form a dense region (core point). • How it works: – Points in dense regions form clusters. – Points in sparse regions or that don’t meet density criteria are considered noise. • Strengths: – Can detect clusters of arbitrary shapes. – Robust to noise. • Weaknesses: – Sensitive to the choice of ε and MinPts.
- 3. OPTICS (Ordering Points to Identify the Clustering Structure) • Concept: OPTICS is an extension of DBSCAN that creates an ordering of points based on their density, allowing for clustering with varying density thresholds. • Key Feature: It does not produce a single clustering, but rather an augmented cluster ordering, making it easier to analyze clusters with varying densities. • How it works: – Points are ordered based on how reachable they are from core points. – No fixed ε; instead, clusters are identified based on a range of density levels. • Strengths: – Can handle clusters of varying densities and sizes. • Weaknesses: – More complex and slower than DBSCAN.
- 4. DENCLUE (DENsity-based CLUstEring) • Clustering based on Density Distribution Functions • Concept: DENCLUE is a density-based method that uses mathematical density functions (such as Gaussian kernels) to model the influence of data points on their surroundings. • Key Features: – Clusters are identified where density is high, and points with low density are outliers. – It builds a density function from data points and identifies local maxima of this function as cluster centers. • How it works: – Each point contributes to a density function, and regions where the density exceeds a threshold form clusters. • Strengths: – Provides a theoretical model for clustering and handles noise well. – Suitable for finding arbitrarily shaped clusters. • Weaknesses: – Requires the careful tuning of parameters like bandwidth for the kernel function.
- 5. Grid-Based Methods • Grid-based clustering methods divide the data space into a finite number of cells that form a grid structure. • The clustering process is then applied to the grid cells rather than the individual data points, which makes them computationally efficient, especially for large datasets. • Key Concept • The data space is partitioned into a grid of cells. • The density of points within each cell is calculated. • Cells with high densities are grouped into clusters. • This reduces the computational complexity as the algorithm operates on the grid instead of the raw data points.
- 6. STING (Statistical Information Grid) • Concept: STING partitions the data space into a hierarchical grid structure where statistical information about data points is stored in each cell. • How it works: – The space is divided into rectangular cells at various resolutions (hierarchical). – Statistical summaries (mean, variance, etc.) are stored at each level. – Cells with high densities are combined to form clusters. • Strengths: – Efficient due to the use of statistical summaries. – Can handle large datasets. • Weaknesses: – Fixed grid structure may not adapt well to varying data densities.
- 7. CLIQUE (Clustering in QUEst) • Concept: CLIQUE is a grid-based clustering algorithm designed for high- dimensional data. It finds dense regions in a subspace of the data and combines these regions to form clusters. • How it works: – The data space is partitioned into an equal number of intervals (grid cells) in each dimension. – Dense cells (cells with a high number of points) are identified in subspaces of the data. – Clusters are formed by combining adjacent dense cells. • Strengths: – Can handle high-dimensional data efficiently. – Automatically finds the best subspaces to cluster the data. • Weaknesses: – Sensitive to the grid size and may produce poor results if the grid is not well-tuned.