Formula
1) ^
β 1=∑ ¿ ¿ ¿ ¿ -> X and Y here is big capital letter;
or ^
β 1= ∑ ¿ ¿ ¿ ¿
or ^
β 1=¿
∑ X i Y i−n( X )(Y )
^ 2
β =¿ ∑ X i −n( x )2
1
or
X =mean value of X;
Y =mean value of Y;
¿¿
¿¿
∑ ¿¿ ¿
2) ^
β 0=Y − ^
β 2 x -> Y and X here is big capital letter, ^β 2 is what you estimated just now
u^
3) σ 2= ∑ i -> variance
2
n−2
∑ u^2i → summation of the square of error term∨Residual ∑ of square (usually question
will give, no need calculate your own)
4) σ =
√ ∑ u^2i
n−2
-> standar error
2
σ
5) Var ( ^
β1¿ = ∑ x -> the x here small capital letter
2
i
2
σ
or Var ( = 1^ ¿
∑ ¿¿¿¿ -> the X here is big capital letter , summation of the square of
your X minus mean value.
2
σ
or Var ( ^
β1¿ = ∑ X i −n¿ ¿ ¿
2
√ √ σ2
2
σ
6) Se ( ^
β1¿ = ∑x 2
i
(small capital x) or ¿
∑ ¿ ¿¿ ¿ (big capital X)
√
2
σ
or ¿
∑ X −n ¿ ¿ ¿
2
i
∑ u^2i
7) R
2
= 1- ∑ ¿ ¿ ¿ -> Y are big capital letter, summation of the square of your Y
minus mean value; ( measure how much variation of Y can be explained by the variation
of X)
or
∑ u^2i
r
2
= 1- ∑ Y −n ¿ ¿ ¿
2
i
Tutorial 2
1.
∑ X i Y i−n( X )(Y )
^ 2
β2 ∑ X i −n( x )2
a) =
830102−22(416.5)(86.65)
= 3919654−22¿ ¿
36128.05
= 103264.5
=0.3499
b) ^
β 1=Y − ^
β2 x
¿ 86.65−(0.3499)(416.5)
= -59.08
c) σ^2= ∑ u^2i
n−2
130.6
= 20
= 6.53
d) Var ( ^
β 2)
2
σ
=
∑ X 2i −n¿ ¿ ¿
6.53
= 3919654−22¿ ¿
=0.000063
e) se ( ^β 2)
= √ 0.000063
=0.00795
2. Table 1 shows the numbers of hours study per day and the marks of 10 KDU
students managed to score for the Investment subject.
Num of
hour
obs study Marks
1 2 65
2 1 55
3 3 68
4 5 86
5 4 88
6 6 92
7 1 50
8 2.5 60
9 3.5 62
10 4.5 90
Given that ∑ u^2i = 10.2
According to the table above, calculate:
a) ^β2
∑ X i Y i−n( X )(Y )
2
∑ X i −n( x )2
=
2545−10( 3.25)(71.6)
= 130.75−10 ¿ ¿
218
= 25.125
= 8.6766
b) ^ β 1=Y − ^ β2 x
¿ ( 71.6 ) −(8.6766)(3.25)
= 43.4
c) σ^2 = ∑ u^2i
n−2
10.2
¿
8
¿ 1.275
d) Var ( ^
β 2)
2
σ
=
∑ X i −n¿ ¿ ¿
2
1.275
= 130.75−10 ¿ ¿
= 0.0507
e) se ( ^
β 2)
= √ 0.0507
=0.2253
f) r2
∑ u^2i
1- ∑ Y −n ¿ ¿ ¿
=
2
i
10.2
= 1- 53522−10 ¿ ¿
= 0.9955
Question 3
In the context of a simple linear regression model, discuss the limitations of using R-
squared as the sole criterion for evaluating model fit. In your answer, explain what R-
squared measures, why it may be misleading, and propose alternative methods for
assessing the quality of a regression model.
Answer:
R-squared (R²), or the coefficient of determination, measures the proportion
of the variation in the dependent variable that is explained by the independent
variable in a regression model. In a simple linear regression, R² is the square
of the correlation coefficient and ranges from 0 to 1. A higher R² generally
indicates that the model explains more of the variance in the dependent
variable.
However, relying solely on R² to assess model quality has several limitations.
First, a high R² does not imply causality. The independent variable may
appear to explain variation in the dependent variable, but the relationship
could be spurious or driven by omitted variables. For example, two unrelated
variables could have high R² due to a common time trend, leading to
misleading conclusions.
Second, R² increases with the addition of more explanatory variables, even if
they are irrelevant. In multiple regression, this can create the illusion of better
model fit. Although this is not a concern in simple regression, it highlights the
need to go beyond R².
Residual analysis helps assess whether the model assumptions (e.g.,
homoscedasticity, linearity, normality of errors) are satisfied.
Standard error of the regression (root MSE) measures the typical
size of residuals and helps assess accuracy.
t-tests and p-values assess the statistical significance of coefficients,
providing insight into whether relationships are likely to hold in the
population.