DEPARTMENT OF STATISTICS

T.Y.B.Sc. Practical No. 5.2.5

LINEAR MODELS
(SOLUTION)

1 103 32
197 36
116 45
X¿ Y=
11 2 25
15 5 32
184 21

6 24 27
'
X X=¿ 24 172 128 | X ' X|=5580
27 128 139

(X’X)-1 = 7524 120 -1572 X’Y = 191

1 120 105 -120 722
5580 -1572 -120 456 912

β^ 0
^
i) Best linear unbiased estimator of β=¿ β^ 1 = (X’X)-1 X1Y
β^
2

β^ 0 16.13978495
β^ 1 = −1.919354839
β^ 5.193548387
2

ii) BLUE of β0 +2 β1- 3 β2= λ β^ where λ ' = [1 2 -3]

'

¿−3.279569889

Y ' Y − ^β X ' Y
'
6435−6433.440861
iii) Estimate of σ 2=¿ σ^ 2= = = 0.519713123
n− p 3

iv) Cov (β^ 1 , β^ 2 ¿

^ ( β^ ) .(X1X)-1σ^ 2
Variance covariance matrix : V
Estimate of COV (β^ 1 , ^β2 ) = x 12 σ^ 2 - 0.01117663

95% confidence interval for β2 : β^ 2 ± t n-p, α/2 √ x 33 σ^ 2 [t3, 0.025 = 3.182]

= 5.193548387 ± t 3 , 0.05 √ 0.042471179

= ( 0.537784738, 5.849312036)
2 E( y i ) =∝+βx i i = 1,2,…….6

2.14 1 180
2.56 1 200
2.92 1 220
y= x=
3.24 1 240
3.44 1 260
3.96 1 280

6 1380
X’X =
1380 324400

1 324400−1380
( X’X)-1 =
42000 −1380 6

18.26
X’Y =
4320.4

^ α^ −0.919238095
i)θ=¿ =
β^ 0.017228571

ii) 95% Confidence Interval for ∝∧β

2
σ =0.04 , Z∝ /2 = 1.96
95% Confidence Interval for ∝=∝ ^ ± Z ∝/2 √ x 11 σ 2
¿(−0.919238095 ±1.089436307)
¿(−2.008674402 ,0.170198211)
95% Confidence Interval for β = β^ ± Z∝ /2 √ x 22 σ 2
= ( 0.017228571 ± 0.004685296
= (0.012543274, 0.021913867).

3 X 1=12.18181818 X 2=6 n = 11, p = 3

1 0 0
1
0 517.63 -46
0 -46 54
X’X =

| X ' X| = 284196.22

.09090 0 0
9
0 .002090105 .001780460
 0 .001780460 .020035

1
(X’X)-1 =
362785.0002

362
(X’Y) = −751.8181
195
 32.909091
^ ( X' X )−1 X’Y = −1.224189
β=
2.568283

^y = β^ 0 + ^β1 ( X1−X1 ) + β^ 2 ( X2 −X2 )

^y = + 32.909091 + (-1.224189) (X 1−X1) + 2.568283 (X 2 - X2)

Least Square Estimate of β0 - β1 + β 2 = λ ' β^ where λ ' = [1 -1 1]

= 36.701563
^ ( λ β^ ) = λ ( X X ) λ σ^
V ' ' ' −1 2

^
2 Y ' Y −β ' X ' Y 13564−13334.27358
σ^ = = = 28.7158025
n− p 8

−1 −1
λ ( X X ) λ=[ 1−11 ]
' ' '
(X X ) 1
-1
1

= 0.109473185
' −1
λ ( X'X ) λ σ^ 2 = 0.524193765

S.E. = √ λ1 ( X ' X)−1 λ σ^ 2 = 0.724012268

4
1 1 1 1 20.2
1 -1 1 -1 8.1
1 1 -1 -1 Y= 9.7
X= 1 -1 -1 1 1.9
1 1 1 1 19.9
1 -1 1 -1 8.3
1 1 -1 -1 10.2
1 -1 -1 1 1.8
| X ' X| = 84

1/8 0 0 0
-1
(X’X) = 0 1/8 0 0
0 0 1/8 0
0 0 0 1/8

80.1
39.9
(X’Y) =
32.9
7.5

10.0125 ^1
w
4.9875 ^2
w
β^ = (X’X)-1 X’Y = 4.1125 =
^3
w
0.9375 ^4
w

' ^'
2 Y Y- β X'Y 1143.53−1143.335
σ^ = = = .04875
n-p 4

ii) 95% Confidence interval for w1 & w2

95% Confidence interval for w1 = ¿ ¿ ± tn-p.∝/2 √ x 11 σ^ 2
= (9.7959, 10.2295)

^ 2 ± tn-p . ∝/2 √ x 11 σ^ 2
95% Confidence interval for w2 = w
= (4.770795572, 5.204201428)

iii) Estimate of w1 – w2 + w3 – w4 = λ ' β^

where λ ’ = [ 1−1 1−1 ]

10.0125
4.9875
∴ λ β^ = [ 1−1 1−1 ]
'
4.1125
0.9375
λ ^β =
'
8.2

^ ( λ ' β^ ) = λ ' ( X ' X )−1 λ σ^ 2

V
1
−1 −1 4 1
λ (X X)
' '
λ = [1/8 -1/8 1/8 -1/8] = =
1 8 2
−1

^ ( λ ' β^ ) = λ ' ( X ' X )−1 λ σ^ 2

V
^ ( λ ' β^ ) = 1 × 0.04875 = 0.024375
V
2
^ '^
V ( λ β) = 0.024375

S.E. = √ V^ (λ ' ^β) = 0.15612495

iv) 95% Confidence Interval for w1 – w2 + w3 – w4 = λ ’β^ ± tn-p , ∝/2 √ V (λ' ^β)
= 8.2 ± (2.776) (0.15612495)
= 8.2 ± 0.433402861
= (7.766597139, 8.633402861)
 1

= 8.2

1
−1
λ σ^ =
' 2
[0.4815] = 0.024375
1
−1

S.E. = √ λ ' ( X' X)-1 λ σ^ 2 = 0.15612495

iv) 95% Confidence Interval for w1 – w2 + w3 – w4

λ ’β^ ± tn-p ∝/2 √ λ ' ( X' X)-1 λ σ^ 2

(7.768597, 8.633402)

v) To test H0 : w1 – w2 + w3 – w4 = 8

test H0 : λ ' β^ = 8 vs H1 : λ ' β^ ≠ 8.

 ' ^
λ β−8
tcal =
√ λ' ( X1 X)−1 λ σ^ 2
8.2−8
=
√6.024375
= 1.281025

tbab = 2.776

tcal < tbab

∴ Do not reject H0 i.e. λ ' β = 8.