Book Notes: Enterprise Analytics - Davenport
My reading notes from the book
Table of Contents
Introduction: The New World of Enterprise Analytics
why now?
powerful computers
increase in data
trained people
need to differentiate products
jobs
tenfold increase from 90 to 2010 (linkedin)
google trends
10x from 2005 to 2012 "analytics"
enterprise analytics
the rise of big data
what is?
too voluminous
too unstructured
IIA and the Research for this Book
International Institute for Analytics: IIA
reserach outputs
research briefs
3-5 page topics
leading-practice briefs
case studies
write-ups of meetings
goal
short documents
initial focus
general enterprise topics
Part I: Overview of Analytics and Their Value
1. What Do We Talk About When We Talk About Analytics?
terms for decision-making from data
70s: decision support systems
80s: executive information systems
80s: online analytical processing (olap)
90s: business intelligence
new: analytics
definition
use of
data
statistical and quantitative analysis
explanatory and predictive models
fact-based management
variations
predictive analytics
data mining
business analytics
web analytics
big-data analytics
Why We Needed a New Term: Issues with Traditional Business Intelligence
bi used
standard reports
answering queries
wikipedia definition
too much verbiage
analytics
contemporary synonym
with more quantitative slant
Three Types of Analytics
descriptive analytics
ex
standard reports
what happened
ad hoc reports
how many, how often?
queries/drill down
what exactly is problem
scorecards
alerts
what actions needed
what happened in past
predictive analytics
statistical modeling
why is it happening
predictive modeling/forecasting
what will happen
use models to predict future from past
prescriptive analytics
randomized testing
what if we try this
optimization
what's the best that can happen
tell you what to do
Where Does Data Mining Fit In?
features
discovery of trends and patterns
automated
intersection of
AI, ML, statistics, database
Business Analytics Versus Other Types
health
health care analytics
informatics
clinical decision support
Web Analytics
new
parts
reporting
A/B testing
Big-Data Analytics
newest
features
too big
too unstructured
too many different sources
2. The Return on Investments in Analytics
Traditional ROI Analysis
roi = ( total value / benefits - total investment ) / total investment
ex
selecting high-potential customers for direct-mail campaign
mine CRM data
send mail to customers who meet a criteria
building model
investment cost: 50 K
expected benefit: 75 K
roi = ( 75 - 50 ) / 50 = 50%
cash flow and roi
assumption: costs and benefits occur at the same time
rarely occurs
credible roi
based on credible business case
The Teradata Method for Evaluating Analytics Investments
process
• Phase 1: Validate business goals and document best-practice usage.
• Phase 2: Envision new capabilities.
• Phase 3: Determine ROI and present findings.
• Phase 4: Communicate.
Phase 1: Validate business goals and document best-practice usage.
includes
strategic business initiatives
progress measures
documenting best practices
reviewing reports, plans, ...
interviewing executives