Lecture Summary 2023
Marketing Strategy Research (not all cases)
Professor Xi Chen
Rotterdam School of Management - Erasmus University Rotterdam
(Includes practice exam)
GOOD LUCK! <3
😀
,Lecture 1 | Introduction
Linking tools to marketing strategy:
- linear regression: market response model (e.g. pricing); it is about making inferences
- conjoint analysis: new product design
- bass model: new product diffusion (e.g. how will you predict the success of a
product)
- cluster analysis: segmentation
- multi-dimensional scaling: positioning
Principles of data-driven marketing: the principles are generic and applicable to almost all
data-driven marketing situations
1. any statistical analysis is to reduce information loss
2. causation cannot be learn directly from data
3. prediction does not care about statistical significance
4. practical usefulness triumphs statistical criteria
,Lecture 2 | Market Response Model
How to predict market response?
Description of a prediction machine: objective is to find
functional relations between input and output → you
put data in the ‘machine’ and get a prediction as output
Linear regression:
y = a + bx
Toy example:
Scatter plot of price vs. sales
Dependent vs. independent variable
Objective: fit the relationship to a ‘line’
Sales = a + bPrice (a = intercept, b = slope/coefficients)
Objective: to find a line given the data points
, What is a good prediction?
Observations vs. predictions
Principle: any statistical analysis is to reduce information loss
How to draw a line
Data point of price = 110 and y = 400, get the difference between actual and predicted
(prediction error)
The objective is to fit the relationship to the line, but what is a good prediction or how do we
draw the line?
- a good prediction captures the information very well and reduces information loss
- choose a line to minimize the differences
Criteria to minimize differences:
- square delta y
- sum up overall points
sales = a + bPrice
a = intercept
b = slope/coefficients
Marketing Strategy Research (not all cases)
Professor Xi Chen
Rotterdam School of Management - Erasmus University Rotterdam
(Includes practice exam)
GOOD LUCK! <3
😀
,Lecture 1 | Introduction
Linking tools to marketing strategy:
- linear regression: market response model (e.g. pricing); it is about making inferences
- conjoint analysis: new product design
- bass model: new product diffusion (e.g. how will you predict the success of a
product)
- cluster analysis: segmentation
- multi-dimensional scaling: positioning
Principles of data-driven marketing: the principles are generic and applicable to almost all
data-driven marketing situations
1. any statistical analysis is to reduce information loss
2. causation cannot be learn directly from data
3. prediction does not care about statistical significance
4. practical usefulness triumphs statistical criteria
,Lecture 2 | Market Response Model
How to predict market response?
Description of a prediction machine: objective is to find
functional relations between input and output → you
put data in the ‘machine’ and get a prediction as output
Linear regression:
y = a + bx
Toy example:
Scatter plot of price vs. sales
Dependent vs. independent variable
Objective: fit the relationship to a ‘line’
Sales = a + bPrice (a = intercept, b = slope/coefficients)
Objective: to find a line given the data points
, What is a good prediction?
Observations vs. predictions
Principle: any statistical analysis is to reduce information loss
How to draw a line
Data point of price = 110 and y = 400, get the difference between actual and predicted
(prediction error)
The objective is to fit the relationship to the line, but what is a good prediction or how do we
draw the line?
- a good prediction captures the information very well and reduces information loss
- choose a line to minimize the differences
Criteria to minimize differences:
- square delta y
- sum up overall points
sales = a + bPrice
a = intercept
b = slope/coefficients
