For Test Flashcards
What is the 5 step framework for linear regression?
Examing the data
Formulating the model
Estimating the model
Validating the model
Making predictions
What is multi collinearity?
When two variables contain the same information
How do we detect multi collinearity?
With the VIF function, >10 means high collinearity
What does r squared imply?
How good the prediction is going to be, the model fit to the data
When R squared is higher this indicates better prediction because of a better fit
What do we use to see to validate a model?
F stats
What is a naive prediction?
A prediction with only intercepts, no other independent vaiables
What is product cannibalization?
New product eats up sales of another brand line
What is a confounding variable?
A third variable that influences both other variables
How do we get rid of confounding variables?
Add them to the regression as a control variable
For what do we use conjoint analysis?
When we want to know what separate attributes should be present in a product
What is the full profiles approach in conjoint analysis?
All attributes are included in the analysis
What is it called when only a part of the attributes are used in the design
Fractional factorial design
What is an orthogonal design of conjoint analysis?
Use a limiting amount profile to answer as much as you can
What is a path worth?
The worth of a specific product to customers
What are the three rules in deciding on the value of pathworths?
Baseline pathworth = 0
Pathworth of insignificant levels = 0
Pathworth of significant levels = coeffients
How is the importance of a partworth defined?
The range of the part worths across the levels, this into a percentage of all ranges combined
What are the three parameters if the bass model?
Innovation propensity p
Market size M
Imitation propensity Q
What is immitation propensity
How susceptible you are to friends decisions, how much do you imitatie them
Innovation propensity what is it?
Own preferences of customers
What us the difference between n(t) and N(t) in the bass model
N is cumulative adoptions
What does a higher p mean?
More early adopters because of more intrinsic motivation
What does higher Q mean
More people adapt each time
What are criteria of good clusters?
Heterogeneity between clusters they need to be different
Homogeneity within clusters you want it to be the same
What are the five steps of cluster analysis?
- Select a distance measure
- Select a clustering procedure
- Decide the number of clusters
- Validate the clustering
- Interpret the clusters
