Predictive Analytics and Data Mining

Index

Note: Page numbers followed by “b”, “f” and “t” indicate boxes, figures and tables respectively

A

AdaBoost model, 158–159, 159f

ANNs, See Artificial neural networks (ANNs)

Anomaly detection

causes

click fraud detection, 331b

credit card transaction fraud monitoring, 331

data errors, 330

distributional assumption, 330

distribution classes, 330

normal data variance, 330

characteristics, 332–333

classification technique, 334

clustering, 334

computer network traffic, 329

density based outlier, 333, 339, 339f–341f

distance based outlier, 333, 333f

data preparation, 336–337, 337f

Detect Outlier (Distances) operator, 337

distance score, 334–335

Euclidean distance, 334

Iris data set, 335, 336f

k-NN classification technique, 334–335

Outlier detection output, 338, 338f