Data Mining Algorithms: Explained Using R

Closing

To be without some of the things you want is an indispensable part of hapiness.

Bertrand Russell

Retrospecting

This book has covered a selection of essential data mining algorithms, mostly originating from machine learning and assuming a machine learning perspective. While both algorithm properties and usage principles are discussed, particular attention has been given to their internal operation mechanisms, explained not only by textual descriptions, equations, and occasionally pseudocode, but also by illustrative and simple R code examples. Apart from algorithms directly used for creating predictive models, techniques for model evaluation, data transformation, and attribute selection have been presented that do not always receive as much attention in data mining practice as they deserve. They may all have significant impact on the final model quality. Regrettably, they also provide opportunities to do things wrong, resulting in producing poor models or overoptimistic performance estimates. The adopted modeling view of data mining was extended even to basic statistics and data transformation. This helps not only to achieve consistency, but also highlight possible pitfalls related to the latter. The above-average space occupied in this book by the discussion of model evaluation techniques and—partially related—methods of incorporating misclassification costs to classification models is motivated by their high practical importance, not always sufficiently recognized.