DrGollerResearchWiki

partitioningclustering

mgoller — 2017-06-11T11:50:48+02:00

Partitioning Clustering Algorithms This section gives a brief overview of the characteristics of partitioning clustering algorithms. Yet as the following paragraphs will show, the process of conceptual clarification is still unfinished. The following survey of partitioning clustering algorithms will point out unclear concepts.

dissertation

mgoller — 2017-06-11T11:40:30+02:00

Introduction Running Example Clustering Overview Distance Functions Partitioning Clustering

distancefunctions - created

mgoller — 2017-06-11T11:34:45+02:00

Expressing Dissimilarity with Distance Function or Distance Matrix A clustering algorithm must be able to determine the similarity of tuples to determine the cluster a tuple is most similar with. Therefore, some kind of measure is required that indicates similarity or dissimilarity of tuples.

clusteringpurposes

mgoller — 2017-06-11T11:06:31+02:00

Purposes of Clustering Clustering for Data Reduction Clustering algorithms scan a set of data and return a set of clusters. The set of data is typically very large in size while the number of clusters is small. Using the circumstance that most clustering algorithms do not store the data that are part of a cluster but a small set of describing features of each cluster, it is possible to use the set of features of a cluster instead of the cluster’s data. In other words, the features that charact…

clusteringoverview

mgoller — 2017-06-11T10:58:31+02:00

The Clustering Problem Clustering is a process that groups a set of objects into several subsets which are called clusters. The goal of the clustering process is to define the clusters in a way that each object and the object that is not more similar to another object are member of the same cluster. In other words, objects within a cluster should be most similar while objects of different clusters should be most different to each other.

start

mgoller — 2017-05-07T22:50:33+02:00

Slowly Changing Measures In a data warehouse system the computation of measures might change over time. Companies that try various ways to handle discount will cause even the simplest measures such as net sales to be subject to change. Complex measures as one can observe them in Big Data settings are more liable to this effect.

diss_runningexample

mgoller — 2016-09-06T23:11:01+02:00

Running Example: Sales in a Data Warehouse For the purpose of illustration this dissertation contains a running example. It is used to demonstrate all concepts presented in this document where it is applicable. Concepts where applying this example does not make sense will be illustrated with their own example

diss_introduction - created

mgoller — 2016-09-04T22:39:12+02:00

Introduction Knowledge discovery in databases (KDD) and data mining offer enterprises and other large organisations a set of methods to discover previously unknown relations among data---which are often called patterns, e.g. han, Frawley, ester. With these patterns enterprises gain deeper insight into their context and are able to use them for better decision-making.

wiki:somekddinstances.png - created

mgoller — 2016-08-28T21:26:11+02:00

wiki:multipleinstances.png - created

mgoller — 2016-08-28T21:24:38+02:00

wiki:iteratingnecessary.png - created

mgoller — 2016-08-28T21:23:23+02:00

wiki:kdd.png - created

mgoller — 2016-08-28T21:22:04+02:00