Problem 2:

Fa
mil
y
siz
Fa Expe Inco e(
mi nditu me(X X2
ly re (Y) 1) )
A 24 11 6
B 8 3 2
C 16 4 1
D 18 7 3
E 24 9 5
F 23 8 4
G 11 5 2
H 15 7 2
I 21 8 3
J 20 7 2

  Expenditure (Y) Income(X1) Family size(X2)

Expenditure (Y) 1 0,884 0,737
Income(X1) 0,884 1 0,867
Family size(X2) 0,737 0,867 1

From the given output, the correlation between the dependent variable (Expenditure (Y)) and the
independent variable (Income (X1)) is 0.884 and the correlation between the dependent variable
(Expenditure (Y)) and the independent variable (Family Size (X2)) is 0.737 which indicates that
the response variable and the independent variables are highly related. Therefore, it is indicated
that the independent variables are highly related with the response variable.

Also, the correlation between the two independent variables (Income (X1)) and (Family
Size(X2)) is 0.867 which implies that they are highly correlated with each other. That is, there is
a multicollinearity present in the regression model.

b.
The fitted regression model is given as follows:
Y^ = β0 + β 1 x 1 + β 2 x 2
Here,
Y^ =expendure
x 1=income
x 2=family ¿ ¿
Use MINITAB to obtain the regression equation of the given data as follows:

  Coefficient Standard t Stat P-value

 s Error
Intercept 3,518651 3,160552 1,113303 0,302342
Annual
Income 2,277618 0,812611 2,80284 0,026416
Family Size -0,41141 1,236033 -0,33284 0,748998
 Y^ =3.52+2.278 x 1−0.411 x 2
The value β 1=2.278 indicates that the Expenditure increases with every additional
Annual
dollars of Income.

The value β 2=−0.411 indicates that the Expenditure decreases with every additional
Family
Size.

Here, β 2=−0.411does not make sense because as the family size increase the expenditure
should also increase.

Also, the correlation between the dependent variable (Expenditure (Y)) and the
independent variable (Family Size (X2)) is 0.727 which indicates that the response
variable and the independent variables are highly related, therefore the value of the
coefficient β 2=−0.411

Therefore, the coefficient β 1=2.278 makes sense but the coefficient make any sense.

does not does not make any sense.

c.
A thumb rule for interpreting the variance inflation factor (VIF) is as follows:

(1) If , then the independent variables are not correlated.

(2) If VIF is between 1 and 5, then the independent variables are moderately correlated.

(3) If VIF is greater than 5, then the independent variables are highly correlated.

From the MINITAB output, the VIF values for the independent variables are 4.02, which
is greater than 1. It indicates that the multicollinearity is present in the model. Therefore,
the multicollinearity problem is present in the model.

The regression model can be modified by removing some of the highly correlated
independent variables, therefore by either removing X1 or X2, the problem of
multicollinearity can be solved.
 Problem 4.

Regression Statistics
0,70564
Multiple R 3
0,49793
R Square 2
0,40379
Adjusted R Square 4
Standard Error 13,9119
Observations 20

ANOVA
Significanc
  df SS MS F eF
1023,71 5,28939
Regression 3 3071,142889 4 1 0,01002
193,541
Residual 16 3096,657111 1
Total 19 6167,8      

Coefficient Standard
  s Error t Stat P-value
0,19190
Intercept -43,1517 31,67054532 -1,36252 7
1,09395
X1 0,371628 0,339709782 7 0,29017
0,24571
X2 0,351523 0,291715879 1,20502 7
1,73187 0,10252
X3 19,1202 11,04019492 2 7

 From the MINITAB output, it is clear that the multiple regression equation is
Y^ =−43.2+0.372 x1 +0.352 x 2+19.1 x 3
. Hence, the value of the regression coefficients is b0 –43.2, b1 is 0.372, b2 is 0.352, and
b3 is 19.1

The appropriate hypotheses for the given model are as follows:

Null hypothesis:
 There is no linear relationship between Y, and X1, X2, and X3.
Alternative hypothesis:

There is a linear relationship between Y, and X1, X2 and X3

Rejection Rule:

If p-value is lesser than the level of significance α =0.05 , then reject the null hypothesis
H0
Use the significance level, α =0.05

From the given output, F data=5.29 and p−value=0.010.

and Here p-value(=0.010)< α =0.05

Therefore, by the rule, the null hypothesis H 0 is rejected.

Hence, the change is statistically significant. That is, there is a linear relationship between
Y, and X1, X2, X3

b.
Obtain the Forecast value.

From the given information, the regression equation is

Y^ =−43.2+0.372 x1 +0.352 x 2+19.1 x 3

. Therefore, the predicted value of 86 for x1, 77 for x2, and 3.4 for x3 is obtained from
the Minitab output in Par (a).

Hence, the forecast value of daily sales is 80.88

c.
Test the hypothesis of the given model.

The appropriate hypotheses for the given model are as follows:

Null hypothesis: H 0=β 1=0

That is, there is no linear relationship between Y and X1

Alternative hypothesis:
H1: β1 ≠ 0
That is, there is a linear relationship between Y and X1
Use the significance level α =0.05

From the given output, t data=1.09 and p-value=0.29

Here, p-value(=0.29) > a (=0.05)
Therefore, by the rule, the null hypothesis is not rejected.

Hence, there is no linear relationship between Y and at 5% level. That is, β 1=0

Obtain the Forecast value.

From the given information, the regression equation is

Y^ =−43.2+0.372 x1 +0.352 x 2+19.1 x 3

Therefore, the predicted value of 86 for , and 3.4 for is obtained from the Minitab output in Part
(a).

Hence, the forecast value of daily sales is 80.88

Null hypothesis: H 0 : β2 =0
That is, there is no linear relationship between Y and X2
Alternative hypothesis: H 0 : β2 ≠ 0
That is, there is a linear relationship between Y and X2

Use the significance level α =0.05

From the given output, t data=1.21 and p-value=0.246

Here, p-value(=0.246) > a (=0.05)
Therefore, by the rule, the null hypothesis is not rejected.

Hence, there is no linear relationship between Y and at 5% level. That is, β 2=0

Null hypothesis: H 0 : β3 =0
That is, there is no linear relationship between Y and X3
Alternative hypothesis: H 0 : β3 ≠ 0
That is, there is a linear relationship between Y and X3

Use the significance level α =0.05

From the given output, t data=1.73 and p-value=0.103

Here, p-value(=0.103) > a (=0.05)
Therefore, by the rule, the null hypothesis is not rejected.

Hence, there is no linear relationship between Y and at 5% level. That is, β 3=0

d.
Therefore, the value for mean leverage is calculated as follows: mean leverage = 4/20=0.2
Hence, the mean leverage value is 0.2

0.4058
0.1772
0.0805
0.2978
0.4137
0.2696
0.0783
0.1031
0.3462
0.2131
0.2006
0.1259
0.1507
0.1830
0.2414
0.2340
0.1491
0.1029
0.1115
Total 0.1157
No, there is no single observation having high leverage points

Problem 5:
a.

Regression Statistics
Multiple R 0,989618791
R Square 0,979345351
Adjusted R Square 0,973444022
Standard Error 12,41167198
Observations 10

ANOVA
Significanc
  df SS MS F eF
Regression 2 51130,10179 25565,05 165,9534 1,27E-06
Residual 7 1078,347209 154,0496
Total 9 52208,449      

  Coefficients Standard Error t Stat P-value

Intercept 7,608306707 8,503171594 0,894761 0,400633
 Y -0,004569928 0,023784989 -0,19213 0,853091
  3,36082E-05 8,92994E-06 3,763536 0,007042

Null hypothesis:

There is no linear relationship between Assets and Accounts and Accounts^2

Alternative hypothesis:

There is a linear relationship between Assets and Accounts and Accounts^2

Use the significance level α =0.05

From the given output, F data=165.95 and p-value=0.0000

Here, p-value(=0.000) > a (=0.05)
Therefore, by the rule, the null hypothesis H 0 is not rejected.

Hence, the model is statistically significant. That is, there is a linear relationship between Assets
and Accounts and Accounts^2.
The coefficient of determination [R-square] is the proportion of variance in the outcome, which
is predicted by the combination of predictor variables. Here, 97% of the variability in the Assets
is explained by Accounts and Accounts^2, predictors of the given model.

b.
The appropriate hypotheses for the given model are as follows:

Null hypothesis: H 0=β 1=0

That is, there is no linear relationship between Assets and Accounts
Alternative hypothesis:
H1: β1 ≠ 0

That is, there is a linear relationship between Assets and Accounts.

There is a linear relationship between Assets and Accounts and Accounts^2

Use the significance level α =0.05

From the given output, t data=−0.19 and p-value=0.853

Here, p-value(=0.853) > a (=0.05)
Therefore, by the rule, the null hypothesis H 0 is not rejected.
Hence, there is no linear relationship between Assets and Accounts at 5% level. That is, β 1=0
Null hypothesis: H 0=β 2=0
That is, there is no linear relationship between Assets and Accounts^2.
Alternative hypothesis:
H 1: β2 ≠ 0

That is, there is a linear relationship between Assets and Accounts^2.

Use the significance level α =0.05

From the given output, t data=3.76 and p-value=0.007

Here, p-value(=0.007) < a (=0.05)
Therefore, by the rule, the null hypothesis H 0 is not rejected.
Hence, there is a linear relationship between Assets and Account^2 .

at 5% level. That is, β 2 ≠ 0

Regression Statistics
Multiple R 0,968272454
R Square 0,937551544
Adjusted R Square 0,929745487
Standard Error 20,18767265
Observations 10

ANOVA
Significanc
  df SS MS F eF
Regression 1 48948,11198 48948,11 120,1056 4,27E-06
Residual 8 3260,337016 407,5421
Total 9 52208,449      

  Coefficients Standard Error t Stat P-value

Intercept -17,12149474 8,778206181 -1,95045 0,086931
Y1 0,083205008 0,007592202 10,95927 4,27E-06

he coefficient of determination [R-square] is the proportion of variance in the outcome, which is

predicted by the combination of predictor variables. Here, 93.8 % of the variability in the assets
is explained by the accounts, predictors of the given model.

From the Minitab output obtained in part (a), it is clear that there is highly correlated with the
independent variables. That is, this implies that the mulitcollinearity problem is presented by
these variables. However, the Minitab output obtained in Part (c), there is a single variable,
which is included in the regression model. Therefore, the coefficient result of the number of
accounts is not the same as the output obtained in Part (a).