Data Mining Functionalities
The kind of patterns that data mining discovers can be classified into
classes via data characterisation and data discrimination. These and other
functionalities will be described below.
·
Characterisation - summarisation of general characteristics of a
target class or data eg. Study characteristics of products whose sales increased
10% during the last year.
·
Classification - the process of finding a set of models (or
functions) that describe and distinguish data classes or concepts, for the
purpose of being able to use the model to predict
the class of objects whose class label is unknown.
·
Data Discrimination - comparison of features in a target class
with those of contrasting classes eg. Comparison between sales that increase 10%
to those whose sales that decreased 30% in the same period
·
Association Analysis - the discovery of association rules showing
attribute values conditions that occur frequently together in a given set of
data eg. Finds the probability that a particular target market within a set
income range will purchase a particular product.
·
Cluster Analysis - analyse data objects without consulting a known
class (unlike classification and prediction).
Objects are grouped based on the principles: 1. Maximise the INTRA-class
similarity and 2. Minimise the INTER-class dissimilarity. Eg. Finding homogenous
subpopulations of customers within a customer database.
·
Outlier Analysis - outliers are data objects that do not comply
with general behaviour or model of the data. Outlier mining for the detection of
rare events eg. Fraud detection. Methods include statistical tests assuming a
distribution model using a distance measures, and deviation based methods.
·
Evolution Analysis - describes trends for objects whose behaviour
changes over time. Distinct features are: 1. Time-series data analysis, 2.
Sequence or periodicity pattern matching, 3. Similarity-based data analysis.
Information on this page are based on Data Mining & Datawarehousing, Farhad Daneshgar, 2001.