Scribd
Upload
83 views7 pages

The Springville Herald Case: Regression

The document analyzes data from the Springville Herald to predict new subscriptions based on telemarketing hours. A regression model was developed with telemarketing hours as the independent variable and new subscriptions as the dependent variable. The regression analysis found that telemarketing hours significantly predicted 80.7% of the variation in new subscriptions. However, the model is only valid for predicting up to 1647.8 telemarketing hours and would provide an inaccurate prediction if given a value above that threshold, such as 2000 hours.

Uploaded by

abikal karki
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
Download as docx, pdf, or txt
83 views7 pages

The Springville Herald Case: Regression

The document analyzes data from the Springville Herald to predict new subscriptions based on telemarketing hours. A regression model was developed with telemarketing hours as the independent variable and new subscriptions as the dependent variable. The regression analysis found that telemarketing hours significantly predicted 80.7% of the variation in new subscriptions. However, the model is only valid for predicting up to 1647.8 telemarketing hours and would provide an inaccurate prediction if given a value above that threshold, such as 2000 hours.

Uploaded by

abikal karki
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1/ 7

The Springville Herald Case

9.1

If we try to forecast new subscription taking only the data of past three months, then data would
not be accurate. Due to less sample forecast we get from sample will be less reliable. Hence, to
get more accurate or more reliable forecast, we need to take more samples.

9.2

The factors other than number of telemarketing hours spent might be useful in predicting the
numbers of new subscription are

1. Other advertisement
2. Economic Factors
3. Legal Factors
4. Competitors telemarketing

9.3

a.

Analysis of Data

Regression

Variables Entered/Removeda
Variables Variables
Model Entered Removed Method
1 Telemarketin
. Enter
gb
a. Dependent Variable: New Subscription
b. All requested variables entered.
Here, we have x as independent variable (telemarketing hours) and y as dependent variable(new
subscription)
Model Summaryb

Adjusted R Std. Error of the


Model R R Square Square Estimate Durbin-Watson
a
1 .899 .807 .799 483.567 1.342

a. Predictors: (Constant), Telemarketing


b. Dependent Variable: New Subscription

Now r2 =80.7%
i.e 86.2% of variation in new subscription is explained by telemarketing hours.

Coefficientsa

Standardized
Unstandardized Coefficients Coefficients Collinearity Statistics

Model B Std. Error Beta t Sig. Tolerance VIF

1 (Constant) -462.320 524.990 -.881 .388

Telemarketing 4.489 .467 .899 9.605 .000 1.000 1.000

a. Dependent Variable: New Subscription

Now,
Testing significance of telemarketing hours at 5% level of significance

H0: B1=0 i.e. Telemarketing hours is not significant.

H1: B1≠0 i.e. Telemarketing hours is significant. (Two-Tailed)

α= 0.05

df= n-p-1

=24-1-1

=22

P-value

0.000
Decision:

Since, p-value (0.000) <α (0.05), H0 is rejected.

Hence, Telemarketing hours is significant.

Now,

Let ^y = b0+b1x be the required regression line.

Here,

b0= -408.089

b1= 4.409

So,

^y = -462.320 +4.489 x is the required equation line.

b.

If x= 1000 hours

Here, or,

or ^y = -462.320 +4.489 x

or ^y = -462.320 +4.489 x (1000)

∴ ^y = 4026.68

Followings are the assumptions to be checked:

Here we have the Histogram of the given data,


Therefore, it is non-normal.

Here,

We have Scatter/Dot diagram,


Therefore, there is homoscedasticity

To check Autocorrelation,

H0: δ=0 i.e.There exists no autocorrelation.

H1: δ≠0 i.e.There exists autocorrelation.

n= 24

α= 0.05

p= 1
Critical values

dL= 1.273

dU= 1.446

Test Statistics

Durbin-Watson (D-W)= 1.342

Decision:

H0 is rejected at 0.05 level of significance.

Hence, there is inconclusive autocorrelation.

Therefore, assumptions are valid.

c.

Here,

x max= 1498

10% of 1498= 149.8

1498+149.8= 1647.8

Therefore, the regression model is valid up to 1647.8.

Again,

Given,

X= 2000 hours.
Since, given value (2000) is greater than valid limit (1647.8), the prediction is inaccurate or
invalid.

You might also like