Linear Regression for Business Statistics

What is Linear Regression?

Linear regression attempts to fit a linear relation between a variable of interest and a set
of variables that may be related to the variable of interest.
Linear Regression for Business Statistics

Simple Regression
A regression with only one explanatory or X variable

Multiple Regression
A regression with two or more explanatory or X variables
Case Study

There is a Sales manager of a toys retail company which sells various kinds of toys in the local market.
This sales manager needs to make some kind of projections about the number of monthly units that
the retail company will be able to sell of this particular toy in the coming half year. In the past she has
been making such projections based on her gut feeling and now wishes to be a little more scientific
about the whole process.
Before we go anywhere!
Linear Regression for Business Statistics
Overview of Regression

1. Modeling
2. Estimation
3. Inference
4. Prediction
Linear Regression for Business Statistics
Overview of Regression

1. Modeling Developing a regression model

2. Estimation
3. Inference
4. Prediction
Linear Regression for Business Statistics
Overview of Regression

1. Modeling Developing a regression model

2. Estimation Using software to estimate the model

3. Inference
4. Prediction
Linear Regression for Business Statistics
Overview of Regression

1. Modeling Developing a regression model

2. Estimation Using software to estimate the model

3. Inference Interpreting the estimated regression model

4. Prediction
Linear Regression for Business Statistics
Overview of Regression

1. Modeling Developing a regression model

2. Estimation Using software to estimate the model

3. Inference Interpreting the estimated regression model

4. Prediction Making predictions about the variable of interest

Linear Regression for Business Statistics
Overview of Regression

1. Modeling Developing a regression model

2. Estimation Using software to estimate the model

3. Inference Interpreting the estimated regression model

4. Prediction Making predictions about the variable of interest

Linear Regression for Business Statistics
1. Modeling Developing a regression model
What variables determine sales?
Linear Regression for Business Statistics
1. Modeling Developing a regression model

Price

Advertising

Promotion
Linear Regression for Business Statistics
1. Modeling Developing a regression model
Linear Regression for Business Statistics
1. Modeling Developing a regression model
 Linear Regression for Business Statistics
1. Modeling Developing a regression model

Variable of interest
“Y” variable
“Dependent” variable
“Response” variable
“Regressed” variable
“L.H.S.” variable
 Linear Regression for Business Statistics
1. Modeling Developing a regression model

Variable of interest “Explanatory” variables

“Y” variable “X” variables
“Dependent” variable “Independent” variables
“Response” variable “Covariate” variables
“Regressed” variable “Regressor” variables
“L.H.S.” variable “R.H.S.” variables
 Linear Regression for Business Statistics
1. Modeling Developing a regression model

Variable of interest “Explanatory” variables Coefficients

“Y” variable “X” variables Parameters
“Dependent” variable “Independent” variables
“Response” variable “Covariate” variables
“Regressed” variable “Regressor” variables
“L.H.S.” variable “R.H.S.” variables
Linear Regression for Business Statistics

1. Modeling Developing a regression model

2. Estimation Using software to estimate the model

3. Inference Interpreting the estimated regression model

4. Prediction Making predictions about the variable of interest

Linear Regression for Business Statistics

1. Modeling Developing a regression model

2. Estimation Using software to estimate the model

3. Inference Interpreting the estimated regression model

4. Prediction Making predictions about the variable of interest

* Data needed for every variable in the regression model *

Linear Regression for Business Statistics

1. Modeling Developing a regression model

2. Estimation Using software to estimate the model

3. Inference Interpreting the estimated regression model

4. Prediction Making predictions about the variable of interest

* Data needed for every variable in the regression model *

Let’s do it in Excel
Linear Regression for Business Statistics
2. Estimation Using software to estimate the model
Linear Regression for Business Statistics

1. Modeling Developing a regression model

2. Estimation Using software to estimate the model

3. Inference Interpreting the estimated regression model

4. Prediction Making predictions about the variable of interest

Linear Regression for Business Statistics
3. Inference Interpreting the estimated regression model
Linear Regression for Business Statistics
3. Inference Interpreting the estimated regression model
Linear Regression for Business Statistics
3. Inference Interpreting the estimated regression model
Linear Regression for Business Statistics
3. Inference Interpreting the estimated regression model
Linear Regression for Business Statistics
3. Inference Interpreting the estimated regression model
Linear Regression for Business Statistics
3. Inference Interpreting the estimated regression model
Linear Regression for Business Statistics
Overview of Regression

1. Modeling Developing a regression model

2. Estimation Using software to estimate the model

3. Inference Interpreting the estimated regression model

4. Prediction Making predictions about the variable of interest

Let’s do it in Excel
Linear Regression for Business Statistics
4. Prediction Making predictions about the variable of interest

For the coming six months, company management is considering

three alternative scenarios for selling this particular toy.

Scenario 1: Price = 9.10$, AdExp = 52,000$, PromExp = 61,000$

Scenario 2: Price = 7.10$, AdExp = 48,000$, PromExp = 57,000$

Scenario 3: Price = 8.10$, AdExp = 50,000$, PromExp = 60,000$

Linear Regression for Business Statistics
4. Prediction Making predictions about the variable of interest

For the coming six months, company management is considering

three alternative scenarios for selling this particular toy.

Scenario 1: Price = 9.10$, AdExp = 52,000$, PromExp = 61,000$

Scenario 2: Price = 7.10$, AdExp = 48,000$, PromExp = 57,000$

Scenario 3: Price = 8.10$, AdExp = 50,000$, PromExp = 60,000$

Which scenario to implement to maximize unit sales?

Linear Regression for Business Statistics
4. Prediction Making predictions about the variable of interest

Scenario 1: Price = 9.10$, AdExp = 52,000$, PromExp = 61,000$

Predicted Sales = 72587.31

Scenario 2: Price = 7.10$, AdExp = 48,000$, PromExp = 57,000$

Predicted Sales = 72892.96

Scenario 3: Price = 8.10$, AdExp = 50,000$, PromExp = 60,000$

Predicted Sales = 74542.75 ≈ 74542

Linear Regression for Business Statistics

Regression is a process that has errors

Linear Regression for Business Statistics

Regression is a process that has errors

Linear Regression for Business Statistics

Regression is a process that has errors

Let´s do this in Excel
Linear Regression for Business Statistics
Linear Regression for Business Statistics
Linear Regression for Business Statistics
Linear Regression for Business Statistics
Linear Regression for Business Statistics

Larger residuals “bad” fit

Smaller residuals ⇒ “good” fit
⇒
Linear Regression for Business Statistics

R-square
[ A “goodness of fit” measure ]
Linear Regression for Business Statistics

R-square
[ A “goodness of fit” measure ]

q Varies from 0 through 1.

q Proportion of variation in the Y variable explained by the regression model.
q Values closer to 1 indicate a good fit.
Least squares line
Linear Regression for Business Statistics
Regression is a process that has errors
q Residuals and Errors.
q R-square: A “goodness of fit” measure.

Why do we have errors in the regression model ?

q
Linear Regression for Business Statistics
Regression is a process that has errors
q Residuals and Errors.
q R-square: A “goodness of fit” measure.

Why do we have errors in the regression model ?

q Omitted variables.
q Functional relationship between the Y and X variables.
q The theory of regression analysis is based on certain
assumptions about these errors.
Case Study
 Case Study

Before even reading the article… what questions should we ask?

 Case Study

Before even reading the article… what questions should we ask?

What is self harm ?

Drinking or excessive exercise was included?
Who responded ?
What about boys?
 Case Study

NAIVE REALISM

TV and social media gossip create naïve realism and they

can lead us badly when we confuse a personal perspective
on the world with some universal truth.
 Case Study

• Government decides to reduce waiting time for doctor (specialist)

appointment

• US ranking rewarded selective undergraduate programs

How can you go around the purpose and achive the KPI?
 Case Study

• Government decides to reduce waiting time for doctor (specialist)

appointment
Doctors decided not to accept new patients

• US ranking rewarded more selective undergraduate programs

University recruited more applicatants and they rejected them
When a MEASURE becomes a TARGET,
it ceases to be a good measure