Académique Documents
Professionnel Documents
Culture Documents
• Analysis of difference
Fresh Coffee
Instant Coffee
Fruit Juice
Packet Tea
LOW Tea Bags HIGH
F2
(CONVENIENCE)
Soft Drinks
Milk
Drinking Chocolate
LOW
Note: F1 (versatility) and F2 (convenience) together account for 78% observed variance
Regression
• Linked to association, but regression seeks to
create a formula to explain the association.
• Two forms; straight line (y = mx + c) or non
linear of several forms
• Regression line plotted by “least squares” and is
the line of “best fit”
• Used extensively in sales forecasting, but
ignores the “point of inflexion” problem seen also
in extrapolation
Baker‟s Box: An example of “best
fit” regression
High Quality
line
Line Of
I Don‟t Expectation
The Golden
Believe Triangle
IT
Credibility
Value Avis
Low Added Country High
Price Price
IT‟S A
Cheap And Rip Off
Cheerful Gullibility
Cheap And
Nasty You Must
Be Joking
Low Quality
Exercise: Expansion at Rotiboy!
• Rotiboy have established that the sales for an
outlet of their business (y) can be defined by the
number of people living within one mile of the
outlet (x) by the line y = 20 + 0.55x, where both
the constant is denominated in „000 baht, and x
as „000 population.
• On the basis of this data what is the likely
weekly revenue (in baht „000) for an outlet
established where there is a local population of
10,000
Multivariate Analysis
• Multivariate regression analysis ( The
Fishbein model of attitude measurement)
• Multivariate Analysis of Variance
(MANOVA)
• Factor analysis: Reduction of complex
variables into smaller number of factors
• Cluster analysis
• CHAID (Chi squared automatic interaction
detection)
Tesco and the use of AID
• Based on customer data collected at POS
• Analysed by AID to establish levels of
product penetration and specific customer
characteristics
• Can isolate groups as small as 5,000 from
a population of 26 million!
• Visit www.dunnhumby.co.uk
Communicating Research
• The communicator
• The message
• The medium
• The Audience
• Feedback
Oral communication
• Stand up, speak up, and shut up. There is
a limit to people‟s concentration
• Graphics need to be simple (not more than
12 figures on a slide). More than this
highlight
• Be aware of your personal body language
• If you use technology, make sure it works!
• Have clear objectives as to what to put
over
Some tips on oral presentations
• Variety; in voice, tone, movement, body
language, all elements of posture, etc. Suit your
body language to the message
• Be sure you can get away with humour. Be very
careful if you use it with an international
audience as it is very cultural!
• Look at your audience, good eye contact
• Do not distribute report beforehand if you want
to be selective in how you take from it.
• Leave time at the end for questions
Presenting statistical results
• Never assume the audience knows as much
about statistics as you do!
• Don‟t try and blind them with statistics
• Remember “there are lies, dam lies, and
statistics” Many are very cynical of them!
• Use statistics to demonstrate data in terms of its
use in making a decision. Applied statistics
rather than statistics for their own sake!
• Explain what they mean!
Written reports
• Length and level of reading are inversely related
• Always provide an annotated executive
summary with guideposts to deeper
explanations as required.
• Make sure the report provides the information
required in order to come to the decision
required
• Use tables, graphs and charts as appropriate to
illustrate data and make points.
• Remember a multicoloured pie chart loses a lot
in black and white!
Specialist AnalyticalTechniques
• CHAID (Chi Squared Automatic Interaction
Detection)
• ACORN / TGI Retail Location
• Market Penetration Analysis
• Sensitivity Analysis
• Resource Allocation