Scribd
Upload
50 views6 pages

HOMEWORK 3 Rishabh Arora

The document reports the results of factor analysis, clustering, and regression modeling performed on survey data about motivations for Florida vacations and preferences regarding ice cream brands. For the Florida vacation data, 3 factors were identified relating to personal relief/enjoyment, experiences, and quality family time. High correlations were found between some survey questions. Clustering identified 3 clusters with different relationships to the factors. For the ice cream brand data, regression identified questions relating to fun, memorable experiences, and best value as influencing recommendations for the Breyers brand, and quality ingredients and enjoyment for private labels. Model fits were better for the original variables than for models using the identified factors. The document also reports on additional analyses including addressing multicollinear

Uploaded by

Rishabh Arora
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
50 views6 pages

HOMEWORK 3 Rishabh Arora

The document reports the results of factor analysis, clustering, and regression modeling performed on survey data about motivations for Florida vacations and preferences regarding ice cream brands. For the Florida vacation data, 3 factors were identified relating to personal relief/enjoyment, experiences, and quality family time. High correlations were found between some survey questions. Clustering identified 3 clusters with different relationships to the factors. For the ice cream brand data, regression identified questions relating to fun, memorable experiences, and best value as influencing recommendations for the Breyers brand, and quality ingredients and enjoyment for private labels. Model fits were better for the original variables than for models using the identified factors. The document also reports on additional analyses including addressing multicollinear

Uploaded by

Rishabh Arora
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 6

10248606

Rishabh Arora

HOMEWORK 3

Q1) Florida vacation data set


Number of factors Names

Q8_1 to Q8_12 3 The factors do make sense


in terms of
motivation.

Factor -1: Personal relief


and enjoyment,
Factor-2: Experience
Factor-3: Quality time
with Family

iii) There is Zero Correlation between the Factors. whereas for the correlation between
the
variables please check the below:
• Q8_1 is highly correlated with Q8_4
• Q8_2 is highly correlated with Q8_5
• Q8_3 is highly correlated with Q8_4
• Q8_4 is highly correlated with Q8_1
• Q8_5 is highly correlated with Q8_9
• Q8_6 is highly correlated with Q8_5
• Q8_7 is highly correlated with Q8_2
• Q8_8 is highly correlated with Q8_6
• Q8_9 is highly correlated with Q8_10
• Q8_10 is highly correlated with Q8_9
• Q8_11 is highly correlated with Q8_12 (Highest Correla/on Factor)
• Q8_12 is highly correlated with Q8_11
10248606
Rishabh Arora

iv)
Number of Clusters Cluster Size

Q8_1 to Q8_12 3 Cluster 1: 149,


Cluster 2: 283,
Cluster 3: 169,

Cluster 1: Far from Factor 1 but similar to Factor 2


Cluster 2: Far from Factor 3 but similar to Factor 1
Cluster 3: Similar to Factor 3

v)

Q2) Ice cream data: Likelihood to recommend the brand Breyers

i. Model fit = 26.8%


ii. Q11_4_3 [Fun], Q11_4_6 [Memorable], Q11_4_7 [Buy as a special treat but not
regularly], Q11_4_10 [Tastes better than more brands] and Q11_4_13 [Best value
for the price]

Q3) Ice cream data: Likelihood to recommend the private labels

i. Model fit = 24.7%


10248606
Rishabh Arora

ii. Q11_7_5 [Premium-Uses better quality ingredients], Q11_7_8 [Good


enough to
keep on hand for regular consumption], Q11_7_12 [Enjoyable], Q11_7_3 [Best
value for the price]

Q4) Yes. People like to recommend Bryers brand for a fun and memorable experience.
It is
recommended as it is an item bought for special occasion and gives customer the best
value
for the price.

Whereas People like to recommend private label due to its better quality ingredients,
regular
consumption and an enjoyable experience. Both Bryers and Private labels provide the
Best
Value for the price for its customers.

Q5) Adjusted R square (model fit) of the regression model run for Q2) with the factors
being
the independent variables is 22.1%, which is less than 26.8%. Similarly for Q3), model
fit of
the regression model with the factors being the independent variables is 22.0% which is
less
than 24.7%.

Question 2 – Model Fit Question 3 – Model Fit

Original Variables 26.8% 24.7%

Factors 22.1% 22.0%

A decrease in model fit is observed because during factor analysis as data is lost.

Q6)

i. Yes, there is an issue of multicollinearity. Because advertising and the interaction of


price and advertising are highly correlated. To correct it we standardize the variables
and then conduct the regression.
10248606
Rishabh Arora

Fig: Without standardizing the variables

Fig: Standardized Variables

ii. Zscore(Ad), Zscore(Price), Zscore(Age) and ZAdPrice are the significant variables
iii. The coefficient of Zscore(Ad) = 3.322 (+ve) and ZAdPrice = 0.216 (+ve).
Moderation
effect of price on advertising = (3.322 + 0.216*ZPrice)ZAd. Therefore as the price is
increased the impact of advertising increases due to the positive moderation effect of
price on advertising.
iv. Adjusted R square = 91.2%

Q7)

i. Q11_4_6 [Memorable], Q11_4_7 [Buy as a special treat but not regularly],


Q11_4_12 [Enjoyable], Q11_4_13 [Best value for the price] have a significant impact
on the dependent variable.

ii. From the es<mates of Ques<on 2 we have found that the logis<c regression model
has found addi<onal Q11_4_12 [Enjoyable] aTribute to be significant but Q11_4_3
[Fun] and Q11_4_10 [Tastes be@er than more brands] are not considered as
significant. The main reason for this can be aTributed to the collapse of [1-6] scoring
of likelihood to recommend as “0” and 7 score as “1”.

Q8)

Number of factors Names


Q9_1 to Q9_12 [Reason 4 The factors do make sense
to buy ice-cream] in terms of
motivation.
Factor-1: Social Delight,
Factor-2: Wide
assortment of Flavours
Factor-3: Personal Delight
(Relaxation
& Enjoyment)
Factor-4: Non-
differentiators
10248606
Rishabh Arora

ii) Specified clusters = 3

Cluster 1: Similar to Factor 3 (People who enjoy ice-cream for relief and enjoyment)
Cluster 2: Far from Factor 3 and not similar to any other Factor.
Cluster 3: Similar to Factor 4 and far from Factor 1

iii)

CLUSTER 1
Type of Regression = Logis<cal
Model Fit: Higher specificity and sensi<vity indicate a beTer fit for the model. Total fit
87.9%
accuracy
Significant Variables: Q11_4_8 [Good enough to keep on hand for consump)on],
Q11_4_10
10248606
Rishabh Arora

[Tastes be<er than most brands], Q11_4_11 [Offers wide variety of flavors],
Q11_4_13
[Best value for the price]

CLUSTER 2

Type of regression = Logis<cal


Model Fit = 83.3% Accuracy.
Significant Variables: Q11_4_6 [Memorable], Q11_4_7 [Buy as a special treat but not
regularly], Q11_4_11 [Offers a wide variety of flavors], Q11_4_12 [Enjoyable],
Q11_4_13
[Best value for the price]

CLUSTER 3

Type of regression = Logis<cal


Model Fit = Model cannot be fi@ed because number of observa)ons is less than or
equal to
number of model parameters

You might also like