Linear Regression

and
Correlation Analysis
2
Scatter Plots and Correlation
A scatter plot (or scatter diagram) is used to
show the relationship between two variables
Correlation analysis is used to measure
strength of the association (linear relationship)
between two variables
Only concerned with strength of the relationship
No causal effect is implied

3
Scatter Plot Examples
y
x
y
x
y
y
x
x
Linear relationships Curvilinear relationships
4
Scatter Plot Examples
y
x
y
x
y
y
x
x
Strong relationships Weak relationships
5
Scatter Plot Examples
y
x
y
x
No relationship
6
Correlation Coefficient
The population correlation coefficient
(rho) measures the strength of the
association between the variables

The sample correlation coefficient r is
an estimate of and is used to
measure the strength of the linear
relationship in the sample
observations
7
Features of and r
Unit free
Range between -1 and 1
The closer to -1, the stronger the
negative linear relationship
The closer to 1, the stronger the positive
linear relationship
The closer to 0, the weaker the linear
relationship
8
r = +.3 r = +1
Examples of Approximate
r Values
y
x
y
x
y
x
y
x
y
x
r = -1 r = -.6 r = 0
9
Calculating the
Correlation Coefficient

=
] ) y y ( ][ ) x x ( [
) y y )( x x (
r
2 2
where:
r = Sample correlation coefficient
n = Sample size
x = Value of the independent variable
y = Value of the dependent variable




=
] ) y ( ) y ( n ][ ) x ( ) x ( n [
y x xy n
r
2 2 2 2
Sample correlation coefficient:
or the algebraic equivalent:
10
Calculation Example
Tree
Height
Trunk
Diameter
y x xy y
2
x
2
35 8 280 1225 64
49 9 441 2401 81
27 7 189 729 49
33 6 198 1089 36
60 13 780 3600 169
21 7 147 441 49
45 11 495 2025 121
51 12 612 2601 144
E=321 E=73 E=3142 E=14111 E=713
11
0
10
20
30
40
50
60
70
0 2 4 6 8 10 12 14
0.886
] (321) ][8(14111) (73) [8(713)
(73)(321) 8(3142)
] y) ( ) y ][n( x) ( ) x [n(
y x xy n
r
2 2
2 2 2 2
=


=


=


Trunk Diameter, x
Tree
Height,
y
Calculation Example
r = 0.886 relatively strong positive
linear association between x and y
12
Introduction to Regression Analysis
Regression analysis is used to:
Predict the value of a dependent variable based on
the value of at least one independent variable
Explain the impact of changes in an independent
variable on the dependent variable
Dependent variable: the variable we wish to
explain
Independent variable: the variable used to
explain the dependent variable
13
Simple Linear Regression Model
Only one independent variable, x
Relationship between x and y is
described by a linear function
Changes in y are assumed to be
caused by changes in x
14
Types of Regression Models
Positive Linear Relationship
Negative Linear Relationship
Relationship NOT Linear
No Relationship
15
x y
1 0
+ + =
Linear component
Population Linear Regression
The population regression model:
Population
y intercept
Population
Slope
Coefficient
Random
Error
term, or
residual
Dependent
Variable
Independent
Variable
Random Error
component
16
Linear Regression Assumptions
Error values () are statistically independent
Error values are normally distributed for any
given value of x
The probability distribution of the errors is
normal
The probability distribution of the errors has
constant variance
The underlying relationship between the x
variable and the y variable is linear
17
Population Linear Regression
Random Error
for this x value

y
x
Observed Value
of y for x
i
Predicted Value
of y for x
i

x y
1 0
+ + =
x
i
Slope =
1
Intercept =
0

i
18
x b b y

1 0 i
+ =
The sample regression line provides an estimate of
the population regression line
Estimated Regression Model
Estimate of
the regression
intercept

Estimate of the
regression slope

Estimated
(or predicted)
y value

Independent
variable
The individual random error terms e
i
have a mean of zero
19
Least Squares Criterion
b
0
and b
1
are obtained by finding the
values of b
0
and b
1
that minimize the
sum of the squared residuals
2
1 0
2 2
x)) b (b (y
) y

(y e
+ =
=

20
The Least Squares Equation
The formulas for b
1
and b
0
are:
algebraic equivalent:

=
n
x
x
n
y x
xy
b
2
2
1
) (

=
2
1
) (
) )( (
x x
y y x x
b
x b y b
1 0
=
and
x b b y

1 0 i
+ =
21
b
0
is the estimated average value of
y when the value of x is zero

b
1
is the estimated change in the
average value of y as a result of a
one-unit change in x
Interpretation of the
Slope and the Intercept
22
Simple Linear Regression Example
A real estate agent wishes to examine the
relationship between the selling price of a home
and its size (measured in square feet)

A random sample of 10 houses is selected
Dependent variable (y) = house price in $1000s
Independent variable (x) = square feet
23
Sample Data for House Price Model
House Price in $1000s
(y)
Square Feet
(x)
245 1400
312 1600
279 1700
308 1875
199 1100
219 1550
405 2350
324 2450
319 1425
255 1700
24
0
50
100
150
200
250
300
350
400
450
0 500 1000 1500 2000 2500 3000
Square Feet
H
o
u
s
e

P
r
i
c
e

(
$
1
0
0
0
s
)

Graphical Presentation
House price model: scatter plot and
regression line
feet) (square 0.10977 98.24833 price house + =
Slope
= 0.10977

Intercept
= 98.248
25
Interpretation of the
Intercept, b
0
b
0
is the estimated average value of Y when the
value of X is zero (if x = 0 is in the range of
observed x values)
Here, no houses had 0 square feet, so b
0
= 98.24833
just indicates that, for houses within the range of sizes
observed, $98,248.33 is the portion of the house price
not explained by square feet
feet) (square 0.10977 98.24833 price house + =
26
Interpretation of the
Slope Coefficient, b
1
feet) (square 0.10977 98.24833 price house + =
b
1
measures the estimated change in the
average value of Y as a result of a one-
unit change in X
Here, b
1
= .10977 tells us that the average value of a
house increases by .10977($1000) = $109.77, on
average, for each additional one square foot of size
27
Least Squares Regression
Properties
The sum of the residuals from the least squares
regression line is 0 ( )
The sum of the squared residuals is a minimum
(minimized )
The simple regression line always passes through the
mean of the y variable and the mean of the x variable
The least squares coefficients are unbiased estimates
of
0
and
1


0 ) ( =

y y
2
)

( y y

28
Explained and Unexplained Variation
Total variation is made up of two parts:
SSR SSE SST + =
Total sum
of Squares
Sum of Squares
Regression
Sum of
Squares Error

=
2
) y y ( SST

=
2
) y

y ( SSE

=
2
) y y

( SSR
where:
= Average value of the dependent variable
y = Observed values of the dependent variable
= Estimated value of y for the given x value
y
y
29
Explained and Unexplained Variation
SST = total sum of squares
Measures the variation of the y
i
values around
their mean y
SSE = error sum of squares
Variation attributable to factors other than the
relationship between x and y
SSR = regression sum of squares
Explained variation attributable to the
relationship between x and y
30
X
i
y
x
y
i
SST = (y
i
- y)
2
SSE = (y
i
- y
i
)
2

.
SSR = (y
i
- y)
2


.
_
_
_
Explained and Unexplained Variation
y
.
y
y
_
y
.
31
The coefficient of determination is the portion
of the total variation in the dependent variable
that is explained by variation in the
independent variable

The coefficient of determination is also called
R-squared and is denoted as R
2

Coefficient of Determination, R
2
SST
SSR
R =
2
1 R 0
2
s s
where
32
Coefficient of determination
Coefficient of Determination, R
2
squares of sum total
regression by explained squares of sum
SST
SSR
R = =
2
Note: In the single independent variable case, the coefficient
of determination is


where:
R
2
= Coefficient of determination
r = Simple correlation coefficient
2 2
r R =
33
R
2
= +1
Examples of Approximate
R
2
Values
y
x
y
x
R
2
= 1
R
2
= 1
Perfect linear relationship
between x and y:

100% of the variation in y is
explained by variation in x
34
Examples of Approximate
R
2
Values
y
x
y
x
0 < R
2
< 1
Weaker linear relationship
between x and y:

Some but not all of the
variation in y is explained
by variation in x
35
Examples of Approximate
R
2
Values
R
2
= 0
No linear relationship
between x and y:

The value of Y does not
depend on x. (None of the
variation in y is explained
by variation in x)
y
x
R
2
= 0
36
Standard Error of Estimate
The standard deviation of the variation of
observations around the regression line is
estimated by
1
=
c
k n
SSE
s
Where
SSE = Sum of squares error
n = Sample size
k = number of independent variables in the model

=
2
) y

y ( SSE
37
House Price
in $000s
(y)
Square Feet
(x)
y
2
x
2
xy
245 1400 60025 1960000 343000
312 1600 97344 2560000 499200
279 1700 77841 2890000 474300
308 1875 94864 3515625 577500
199 1100 39601 1210000 218900
219 1550 47961 2402500 339450
405 2350 164025 5522500 951750
324 2450 104976 6002500 793800
319 1425 101761 2030625 454575
255 1700 65025 2890000 433500
Sums 2865 17150 853423 30983750 5085975
Means 286.5 1715
Example: House Prices
38
House Price
in $000s
(y)
Square Feet
(x)
y
2
x
2
xy
Sums 2865 17150 853423 30983750 5085975
Means 286.5 1715
Example: House Prices

=
n
x
x
n
y x
xy
b
2
2
1
) (
x b y b
1 0
=
b
1
= 0.1098
b
0
= 98.25
(sq.ft.) 0.1098 98.25 price house + =
Estimated Regression Equation:
39
House Price
in $1000s
(y)
Square Feet
(x)
245 1400
312 1600
279 1700
308 1875
199 1100
219 1550
405 2350
324 2450
319 1425
255 1700
(sq.ft.) 0.1098 98.25 price house + =
Estimated Regression Equation:
Example: House Prices
Predict the price for a house
with 2000 square feet
40
317.85
0) 0.1098(200 98.25
(sq.ft.) 0.1098 98.25 price house
=
+ =
+ =
Example: House Prices
Predict the price for a house
with 2000 square feet:
The predicted price for a house with 2000
square feet is 317.85($1,000s) = $317,850
41
The Standard Deviation of the
Regression Slope
The standard error of the regression slope
coefficient (b
1
) is estimated by

=
n
x) (
x
s
) x (x
s
s
2
2

b
1
where:
= Estimate of the standard error of the least squares slope

= Sample standard error of the estimate
1
b
s
2 n
SSE
s

=
42
Comparing Standard Errors
y
y y
x
x
x
y
x
1
b
s small
1
b
s large
c
s small
c
s large
Variation of observed y values
from the regression line
Variation in the slope of regression
lines from different possible samples
43
Inference about the Slope:
t Test
t test for a population slope
Is there a linear relationship between x and y?
Null and alternative hypotheses
H
0
:
1
= 0 (no linear relationship)
H
1
:
1
= 0 (linear relationship does exist)
Test statistic





1
b
1 1
s
b
t

=
2 n d.f. =
where:
b
1
= Sample regression slope
coefficient

1
= Hypothesized slope
s
b1
= Estimator of the standard
error of the slope
44
House Price
in $1000s
(y)
Square Feet
(x)
245 1400
312 1600
279 1700
308 1875
199 1100
219 1550
405 2350
324 2450
319 1425
255 1700
(sq.ft.) 0.1098 98.25 price house + =
Estimated Regression Equation:
The slope of this model is 0.1098
Does square footage of the house
affect its sales price?
Inference about the Slope:
t Test
45
Inferences about the Slope:
t Test Example
H
0
:
1
= 0
H
A
:
1
= 0
Test Statistic: t = 3.329
There is sufficient evidence that
square footage affects house
price
From Excel output:
Reject H
0
Coefficients Standard Error t Stat P-value
Intercept 98.24833 58.03348 1.69296 0.12892
Square Feet 0.10977 0.03297 3.32938 0.01039
1
b
s
t b
1
Decision:
Conclusion:
Reject H
0
Reject H
0
o/2=.025
-t
/2
Do not reject H
0
0

t
/2
o/2=.025
-2.3060 2.3060 3.329
d.f. = 10-2 = 8
46
Regression Analysis for Description
Confidence Interval Estimate of the Slope:
Excel Printout for House Prices:
At 95% level of confidence, the confidence interval for
the slope is (0.0337, 0.1858)
1
b /2 1
s t b
o

Coefficients Standard Error t Stat P-value Lower 95% Upper 95%

Intercept 98.24833 58.03348 1.69296 0.12892 -35.57720 232.07386
Square Feet 0.10977 0.03297 3.32938 0.01039 0.03374 0.18580
d.f. = n - 2
47
Regression Analysis for Description
Since the units of the house price variable is
$1000s, we are 95% confident that the average
impact on sales price is between $33.70 and
$185.80 per square foot of house size
Coefficients Standard Error t Stat P-value Lower 95% Upper 95%
Intercept 98.24833 58.03348 1.69296 0.12892 -35.57720 232.07386
Square Feet 0.10977 0.03297 3.32938 0.01039 0.03374 0.18580
This 95% confidence interval does not include 0.
Conclusion: There is a significant relationship between
house price and square feet at the .05 level of significance
48
Case Study
The following data gives the Sales & Net Profit for some of
the top Auto-makers during the quarter July-Sep 2006
(Rs. Crores)
Company Sales Net Profit
Tata Motors 6484.8 466.0
Hero Honda 2196.5 224.2
Bajaj Auto 2444.7 345.4
TVS Motor 1032.9 35.1
Bharat Forge 461.6 63.4
Ashok Leyland 1635.8 94.7
M&M 2365.5 200.6
Maruti Udyog 3426.5 315.7
49
Case Study
Fit a regression equation for Net Profit on Sales
Maruti expects its Sales to go up to 3800 crores by the
introduction of a new segment A2 model. Find out a 90%
confidence interval for the expected net profit.
Find out the Coefficient of Determination & the
Correlation between Sales & Net Profit. Does this
indicate a strong association?
An earlier study done in 2001 indicated that for a sale of
100 the net profit was 80. Does the current data indicate
a change in this trend. Test at 5% level of confidence.

50
Levin Rubin Computations for Ch 12 problems
PROBLEM n X Y X
2
Y
2
Mean X Mean Y XY
12-16 10 37.2 75.5 147.18 597.03 3.72 7.55 295.95
12-17 5 28 25 158.5 175 5.6 5 131
12-18 8 340 5500 15500 3830800 42.5 687.5 227200
12-19 8 120 228 2100 8884 15 28.5 2580
12-20 8 28 260 128 9320 3.5 32.5 1047
12-21 10 34 1371 122.62 201121 3.4 137.1 4954
12-22 7 146 48 3550 352 20.857 6.857 1101
12-23 8 104.05 1630 159.70 349814 13 203.75 17746.6
12-24 13 185.8 384 3177.12 11755.6 14.29 29.54 5080.27
12-31 13 42 117.6 158 1187.84 3.23 9.046 337
12-32 8 23 66 99 700 2.875 8.25 246
12-34 12 128.7 306.95 1633.41 8838.59 10.725 25.58 3786.64
12-37 11 16 1071 23.88 106285 1.4545 97.3636 1591.8