Data Mining Techniques
Classifying data mining techiques has been allways a sensitive subject. There are dozens of classifications of data mining with classes,
sub-classes and sub-sub-classes. And sometimes a particular class may have dozens of names.
One can argue that particular methods are not data mining techiques but pure statistical concepts. So I'll never brag about being able to
draw the complete picture. But I hope one can have at least an idea about the data mining methods by looking at the classification below.
I should also add that it refers only to the "predictive" part of the data mining.
The other part, the "descriptive" one, it relates more to statistics
rather than data mining (correlations, ANOVA, Ztest, histograms, etc). This is not to say that I don't consider regression a statistical
concept. I do, but, again, the border between statistics and
data mining is sometimes blurred. The science of statistics has itself a well rounded set of predictive methods
that were borrowed by data mining and never returned...
- Association Analysis
- With Candidate Generation
Algorithms: APRIORI
- Without Candidate Generation
Algorithms: RELIM, FP-GROWTH
- Classification
- Decision Trees
Algorithms: CHART, CAID, C4.5
- Artificial Neural Networks
Algorithms: SLP, COHONEN, MNP
- Bayesian Classification
- K-Nearest Neighbors
Algorithms: PEBLS
- Support Vector Machine
- Genetic Algorithm
- Clustering
- Hierarchical
Agglomerative Algorithms: HACM, SLINK, COBWEB, BIRCH, CURE, ROCK, CHAMELEON
Divisive Algorithms:
- Partitional
Algorithms: K-MEANS, CLARA, CLARANS, PAM
- Density Based
Algorithms: DBSCAN, OPTICS
- Regression
- Linear Regression
- Non-Linear Regression
- Logical Regression
|
|