ñòð. 132 |

dimension tables, 502â€“503

patterns

discussed, 31

meaningful discoveries, 56

levels of, 475

prediction, 45

logical schema, 478

untruthful learning sources, 45â€“46

metadata, 483â€“484, 491

peg values, 236

operational summary data, 477

penetration, proportion, 203

physical schema, 478

percent variations, 105

reporting requirements, 495â€“496

perceptrons, defined, 212

transaction data, 476â€“477

Index 635

distribution and, 135

performance, classification, 12

hazards, 394â€“396

physical schema, OLAP, 478

statistics, 133â€“135

pilot projects, 598

probation periods, 518

planar graphs, 323

problem management

planned processes, proof-of-concept

data transformation, 56â€“57

projects, 599

identification, 43

platforms, data mining, 527

lift ratio, 83

point of maximum benefit, 101

profiling as, 53â€“54

point-of-sale data

rule-oriented problems, 176

association rules, 288

variable selection problems, neural

scanners, 3

networks, 233

as useful data source, 60

products

population diversity, 178

clustering by usage, market based

positive ratings, voting, 284

analysis, 294â€“295

postcards, as communication

co-occurrence of, 299

channel, 89

hierarchical categories, 305

potential revenue, behavior-based

information as, 14

variables, 583â€“585

introduction, planning for, 27

precision measurements, classification

product codes, as categorical

codes, 273â€“274

value, 239

preclassified tests, 79

product-focused businesses, 2

predictions

taxonomy, 305

accuracy, 79

profiling

association rules, 70

business goals, formulating, 605

business goals, formulating, 605

collaborative filtering, 283â€“284

collaborative filtering, 284â€“285

data transformation, 57

credit risks, 113â€“114

decision trees, 12

customer longevity, 119â€“120

demographic profiles, 31

data transformation, 57

descriptive, 52

defined, 52

directed, 52

directed data mining, 57

examples of, 54

errors, 191

gender example, 12

future behaviors, 10

new customer information, 283

historical data, 10

overview, 12

model sets for, 70â€“71

predication versus, 52â€“53

neural networks, 215

as problem management, 53â€“54

patterns, 45

survey response, 53

prediction task examples, 10

profitability

profiling versus, 52â€“53

marketing campaigns, 100â€“104

response, MBR, 258

proof-of concept projects, 599

uses for, 54

results, assessing, 85

probabilities

projective visualization (Marc

calculating, 309

Goodman), 206â€“208

class labels, 85

636 Index

proof-of-concept projects

planning, 27

expectations, 599

profitability, 100â€“104

identifying, 599â€“601

response modeling, 96â€“97

implementation, 601â€“605

types of, 111

propensity

up-selling, 115â€“116

categorical variables, 242

messages, selecting appropriate,

propensity-to-respond score, 97

89â€“90

proportion ranking, 88â€“89

converting counts to, 75â€“76 roles in, 88

difference of proportion targeting, 88

chi-square tests versus, 153â€“154 time dependency and, 160

statistical analysis, 143â€“144

prospective customer value, 115

penetration, 203

prototypes, proof-of-concept

standard error of, statistical analysis,

projects, 599

139â€“141 pruning, decision trees

proportional hazards

C5 algorithm, 190â€“191

Cox, 410â€“411

CART algorithm, 185, 188â€“189

discussed, 408

discussed, 184

examples of, 409

minimum support pruning, 312

limitations of, 411â€“412

stability-based, 191â€“192

proportional scoring, census data, public records, house-hold level

94â€“95 data, 96

prospecting

publications

advertising techniques, 90â€“94

Building the Data Warehouse (Bill

ñòð. 132 |