<<

. 129
( 137 .)



>>


information brokers, supermarket

considerations, 51

chains as, 15“16
decision-making process, 50“51

information gain, entropy, 178“180
generating, 51

information technology, data
market basket analysis, 51

transformation, 58“60

null hypothesis, statistics and,

as products, 14

125“126
recommendation-based businesses,

I 16“17
IBM, relational database management Inmon, Bill (Building the Data
software, 13
Warehouse), 474

ID and key variables, 554
input columns, 547

ID3 (Iteractive Dichotomiser 3), 190
input layer, free-forward neural

identification
networks, 226

columns, 548
input variables, target fields, 37

customer signatures, 560“562
inputs/outputs, neural networks, 215

good prospects, 88“89
insourcing data mining, 524“525

problem management, 43
insurance claims, classification, 9

proof-of-concept projects, 599“601
interactive systems, response times, 33

identified versus anonymous Internet resources

transactions, association rules, 308
customer response to marketing

identity distance, distance function, 271
campaigns, tracking, 109

ignored columns, 547
RuleQuest, 190

images, binary data, 557
U.S. Census Bureau, 94

imperfections, in data, 34
interval variables, 549, 552

implementation
interviews

neural networks, 212
business opportunities,

proof-of-concept projects, 601“605
identifying, 27

implicit parallelism, 438
proof-of-concept projects, 600

in-between relationships, customer
intrinsic information, splits, decision

relationships, 453
trees, 180

income, house-hold level data, 96
introduction, of products, 27

630 Index


intuition, data exploration, 65
case study, 343“346

involuntary churn, 118“119, 521
classification, 9

item popularity, market based
discussed, 321

analysis, 293
fax machines, 337“341

item sets, market based analysis, 289
graphs

Iterative Dichotomiser 3 (ID3), 190
acyclic graphs, 331

communities of interest, 346

K cyclic, 330“331

key and ID variables, 554
data as, 340

KDD (knowledge discovery in
directed graphs, 330

databases), 8
edges, 322

Kimball, Ralph (The Data Warehouse
graph-coloring algorithm, 340“341

Toolkit), 474
Hamiltonian path, 328

Kleinberg algorithm, link analysis,
nodes, 322

332“333
planar graphs, 323

K-means clustering, 354“358
traveling salesman problem,

knowledge discovery in databases
327“329

(KDD), 8
vertices, 322

Kolmogorov-Smirnov (KS) tests, 101
hubs, 332“334

Kleinberg algorithm, 332“333

L root sets, 333

large-business relationships, customer
search programs, 331

relationship management, 3“4
stemming, 333

leaf nodes, classification, 167
weighted graphs, 322, 324

learning
linkage graphs, 77

opportunities, customer interactions,
lists, ordered and unordered, 239

520“521
literature, market research, 22

supervised, 57
logarithms, data transformation, 74

training techniques as, 231
logical schema, OLAP, 478

truthful sources, 48“50
logistic methods, box diagrams, 200

unsupervised, 57
long form, census data, 94

untruthful sources, 44“48
long-term trends, 75

life stages, customer relationships,
lookup tables, auxiliary information,

455“456
570“571

lifetime customer value, customer
loyalty

relationships, 32
customers, 520

lift ratio
loyalty programs

comparing models using, 81“82
marketing campaigns, 111

lift charts, 82, 84
welcome periods, 518

problems with, 83
luminosity, 351

linear processes, 55

M
linear regression, 139

link analysis
mailings

authorities, 333“334
marketing campaigns, 97

candidates, 333
non-response models, 35

Index 631


as statistical analysis

marginal customers, 553

acuity of testing, 147“148

market based analysis

confidence intervals, 146

differentiation, 289

proportion, standard error of,

discussed, 287

139“141
geographic attributes, 293

results, comparing, using confi­
item popularity, 293

dence bounds, 141“143
item sets, 289

sample sizes, 145

market basket data, 51, 289“291

targeted acquisition campaigns, 31

marketing interventions, tracking,

types of, 111

293“294

up-selling, 115“116

order characteristics, 292

usage stimulation, 111

products, clustering by usage,

<<

. 129
( 137 .)



>>