Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

MKTG 376 Assignment 2

Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 5
At a glance
Powered by AI
This document discusses using linear and log linear regression models to analyze the sales of Tropicana orange juice based on price variables and display.

The regression models analyze the sales of Tropicana based on its own price, competitors' prices (Minute Maid and Domick's), and a display dummy variable.

The coefficient of the price variable in the linear regression estimates that as the price of Tropicana increases by $1/week, sales decrease by around $21,718 on average, holding all else constant.

MKTG 376

Assignment 2: Regression
Due: February 19, at the Start of Class

Data Overview
This is a sample of 116 weekly sales of Tropicana oranges juice, prices of Tropicana, Minute
Made, Domick’s orange juice, and dummy variables for feature and display. We want to predict
sales of Tropicana give its price, its competitor’s prices and whether it is on display.

The data has the variables:

Variable Description

Week The number of week in which data was collected


SalesTrop Sales of Tropicana for that week
PriceTrop Price of Tropicana for that week
PriceMM Price of Minute Maid for that week
PriceDom Price of Domick for that week
Display Dummy to indicate if Tropicana was on display in that week

Objective: Learn how


 estimate regression models
 transform variables

Step 1: Open Tropicana Data in Excel


Step 2: Run a Multiple Linear Regression (15 points)

Estimate the following model (full model):

SalesTrop = a + b1×PriceTrop + b2×PriceMM + b3×PriceDom + b4×Display

Data Analysis  Regression


Select the appropriate data for Input Y Range and Input X Range

a. Make a scatter plot: x-axis: Predicted Sales vs. y-axis: Residuals. Copy and paste here.
Predicted Sales vs. Residuals
60000

50000

40000

Residuals 30000

20000

10000

0
-10000 0 10000 20000 30000 40000 50000 60000 70000
-10000

-20000

-30000
Predicted Sales
a.

b. Is the regression overall significant? (Hint: When Significance F< 0.05, the regression is
overall significant)
Yes this regression is overall significant because the Significance F is 7.94E-26.

c. Find R2 and interpret its meaning in the words of the problem.


We are 66.94% of the way to perfectly predicting the sales of Tropicana given its competitor
prices (Minute Maid, Domick’s juice) and whether it is on display.

d. What is your standard error and what does it mean in words?


The sales of Tropicana weekly are off by an average $1,1764.84

e. What are the coefficients and its corresponding p-value of the regression? Fill in the blank.

  Coefficients P-value
Intercept 55959.37 1.52E-08
PriceTrop -21718.04 4.33E-16
PriceMM 8881.79 0.00
PriceDo 915.2493 0.763558
m
Display 19689.74 9.14E-09
f. Given the regression output summary, which independent variables are significant to predict
the Sales of Tropicana? (Hint: When p-value < 0.05, the coefficient is significant)

The price of Minute Maid, Tropicana, and the dummy variables for feature and display are
all significant variables because their p-value is < 0.05.

g. What is the coefficient of PriceTrop? Interpret, using words, what your coefficient of
PriceTrop estimates mean.
As the price of Tropicana increases by $1/week, the sales of Tropicana decreases by
$21,718.042 sales/week, on average all else constant

h. What is the coefficient of PriceMM? Is this coefficient positive? What is the meaning of this
coefficient being positive?
As the price of Minute Maid increases by $1/week, the sales of Tropicana increases by
$8881.79 sales/week, on average all else constant
Step 3: Create some Ln Variables and Re-run Regression (10 points)

Now we need to estimate the following model, where we drop all insignificant independent
variables, and use the ln(SalesTrop), ln(PriceTrop) and ln(PriceMM) instead of SalesTrop,
PriceTrop and PriceMM.

ln(SalesTrop) = a + b1×ln(PriceTrop) + b2×ln(PriceMM) + b3×Display

To do this we need to create the transformed variable.


1. Create new column (new variable) in the data called LnSales using the Excel function
=ln(SalesTrop)
2. Create new column (new variable) in the data called LnPriceTrop using the Excel
function =ln(PriceTrop)
3. Create new column (new variable) in the data called Ln PriceMM using the Excel
function =ln(PriceMM)

Estimate the linear regression model above.

Hint: you should get regression summary output as below:


Coefficient Standard Lower Upper
  s Error t Stat P-value 95% 95%
0.22371 52.0796 11.2075
Intercept 11.65081 1 9 2.46E-80 6 12.09407
LnPriceTro 0.17156
p -2.604788 5 -15.1825 5.96E-29 -2.94472 -2.26485
0.5596820 0.17789 3.14614 0.00211 0.20720
LnPriceMM 6 4 6 9 7 0.912157
0.6379432 0.09294 6.86357 0.45378
Display 7 6 5 3.92E-10 2 0.822104

a. Make a scatter plot: x-axis: Predicted Sales vs. y-axis: Residuals. Copy and paste here.
Presidcted Ln Sales vs. Residuals
1.5

0.5
Residuals

0
0 2 4 6 8 10 12
-0.5

-1

-1.5
Presdicted Ln Sales
b. What is the value of the coefficient of ln(PriceTrop)? What is the own price elasticity of
demand? Is the coefficient of ln(PriceTrop) negative or positive? Why?

The value of ln(PriceTrop) is -2.605%. The price of elasticity of Tropicana is as the price
increases by 1%, this leads to 2.605% decrease in sales. The coefficient is negative because
there is a negative correlation between the ln(priceTropicana) and the ln(salesTropicana).

c. What is the value of the coefficient of ln(PriceMM)? What is the cross price elasticity of
demand? Is the coefficient of ln(PriceMM) is negative or positive? Why?

The value of ln(PriceMM) is 0.5597%. The cross price elasticity of Minute Maid is as the
price increase by 1%, this will lead to a 0.5597% increase in sales. The coefficient is
positive because there is a positive correlation between the ln(priceMM) and the
ln(salesTropicana).

d. What is value of the coefficient of Display? Is this coefficient positive or negative? How
would you interpret the sign (positive or negative) of this coefficient?
If Tropicana is displayed, its weekly sales increases by $0.638 sales/week. This is a positive
coefficient because there is a positive correlation between whether it is on display and the
ln(salesTropicana).

You might also like