15714159
Descripción
Fichas por Tyson Mcleod, actualizado hace más de 1 año
|
Creado por Tyson Mcleod
hace alrededor de 6 años
|
|
| Pregunta | Respuesta |
| What is clustering? | Finding groups of objects such that objects in one group will be similar to one another and different from objects in other groups |
| Partitional clustering | Data objects in non-overlapping subsets (clusters) such that each object is in exactly one subset. |
| Hierarchical clustering | A set of nested clusters organized as a hierarchical tree |
| Clustering algorithms (3 used in this course) | K - means (partitional) Density - based clustering Hierarchical clustering |
| Clustering distinctions 1. Exclusive versus non-exclusive 2. Fuzzy versus non-fuzzy 3. Partial versus complete 4. Heterogeneous versus homogeneous | 1. non-exclusive: points may belong to multiple clusters 2. a point belongs to every cluster with weight between 0 and 1. weights must sum to 1 3. partial cluster: only want to cluster some of the data 4. clusters of widely different sizes, shapes and densities |
| Centroid | (typically) The mean of the points in the cluster |
| K - means complexity | O(n*K*I*d) n = number of points K = number of clusters I = number of iterations d = number of attributes |
| Well-separated cluster | Every point in a cluster is closer to every other point in the cluster, than to any point not in the cluster |
| Center-based cluster | Every point in the cluster is close to the center of the cluster (often the centroid) |
| Contiguity- based cluster | A point in a a cluster is closer to one or more other points in the cluster than to any point not in the cluster |
| Density-based cluster | A cluster is a dense region of points separated by low-density regions from other regions of high density |
| Shared property clusters (conceptual clusters) | Clusters that share some common property or represent a particular concept |
| Pre-processing and post-processing (clustering) | Pre-processing: normalize the data and eliminate outliers Post-processing: 1. eliminate small clusters that may represent outliers. 2.Split loose clusters with high SSE. 3.Merge clusters with low SSE. |
| Strengths of hierarchical clustering | * Desired number of clusters can be obtained by cutting the dendogram. * Meaningful taxonomies e.g. (animal kingdom) |
| Two main types of hierarchical clustering | Agglomerative Divisive |
| Cluster similarity - MIN or Single Link | Similarity of two clusters is based on the two most similar (closest) points in the different clusters |
| Cluster similarity - MAX or Complete Link | Similarity of two clusters is based on the two most different (distant) points in the different clusters |
| Cluster similarity - Group average | Proximity of two clusters is the average of pairwise proximity between points in two clusters |
| Cluster similarity - Ward's method | Similarity of two clusters is based on the increase in squared error when two clusters are merged. Less susceptible to noise. |
| DBSCAN - density | Density = the number of points within a specified radius Eps |
| DBSCAN - MinPts | A point is a core point if it has more than a specified number of points (MinPts) within Eps |
| DBSCAN - Border | A border point has fewer than MinPts within Eps, but is in the neighborhood of a core point. |
| DBSCAN - Noise | A noise point is any point that is not a core point or a border point |
| DBSCAN - pros and cons | Pros: cluster of arbitrary shape, robust to noise, does not need a priori k Cons: required connected regions of sufficiently high density, data sets with varying densities are problematic |
¿Quieres crear tus propias Fichas gratiscon GoConqr? Más información.