Title: Linear Regression
1Linear Regression
- Hypothesis testing and Estimation
2- Assume that we have collected data on two
variables X and Y. Let - (x1, y1) (x2, y2) (x3, y3) (xn, yn)
- denote the pairs of measurements on the on two
variables X and Y for n cases in a sample (or
population)
3The Statistical Model
4- Each yi is assumed to be randomly generated from
a normal distribution with - mean mi a bxi and
- standard deviation s.
- (a, b and s are unknown)
5The Data The Linear Regression Model
- The data falls roughly about a straight line.
unseen
6The Least Squares Line
- Fitting the best straight line
- to linear data
7- Let
- Y a b X
- denote an arbitrary equation of a straight line.
- a and b are known values.
- This equation can be used to predict for each
value of X, the value of Y. - For example, if X xi (as for the ith case) then
the predicted value of Y is -
8- The residual
- can be computed for each case in the sample,
- The residual sum of squares (RSS) is
- a measure of the goodness of fit of the line
- Y a bX to the data
-
9- The optimal choice of a and b will result in the
residual sum of squares -
- attaining a minimum.
- If this is the case than the line
- Y a bX
- is called the Least Squares Line
-
10- The equation for the least squares line
- Let
-
11Computing Formulae
12- Then the slope of the least squares line can be
shown to be
13- and the intercept of the least squares line can
be shown to be
14- The residual sum of Squares
15- Estimating s, the standard deviation in the
regression model
This estimate of s is said to be based on n 2
degrees of freedom
16Sampling distributions of the estimators
17- The sampling distribution slope of the least
squares line
It can be shown that b has a normal distribution
with mean and standard deviation
18has a standard normal distribution, and
has a t distribution with df n - 2
19- (1 a)100 Confidence Limits for slope b
ta/2 critical value for the t-distribution with n
2 degrees of freedom
20The test statistic is
- has a t distribution with df n 2 if H0 is
true.
21Reject
df n 2
This is a two tailed tests. One tailed tests are
also possible
22- The sampling distribution intercept of the least
squares line
It can be shown that a has a normal distribution
with mean and standard deviation
23has a standard normal distribution and
has a t distribution with df n - 2
24- (1 a)100 Confidence Limits for intercept a
ta/2 critical value for the t-distribution with n
2 degrees of freedom
25The test statistic is
- has a t distribution with df n 2 if H0 is
true.
26Reject
df n 2
27Example
28The following data showed the per capita
consumption of cigarettes per month (X) in
various countries in 1930, and the death rates
from lung cancer for men in 1950. TABLE Per
capita consumption of cigarettes per month (Xi)
in n 11 countries in 1930, and the death
rates, Yi (per 100,000), from lung cancer for men
in 1950. Country (i) Xi Yi Australia 48 18
Canada 50 15 Denmark 38 17 Finland 110 35 Great
Britain 110 46 Holland 49 24 Iceland 23 6 Norw
ay 25 9 Sweden 30 11 Switzerland 51 25 USA 130
20
29(No Transcript)
30Fitting the Least Squares Line
31Fitting the Least Squares Line
First compute the following three quantities
32Computing Estimate of Slope (b), Intercept (a)
and standard deviation (s),
33- 95 Confidence Limits for slope b
0.0706 to 0.3862
t.025 2.262 critical value for the
t-distribution with 9 degrees of freedom
34- 95 Confidence Limits for intercept a
-4.34 to 17.85
t.025 2.262 critical value for the
t-distribution with 9 degrees of freedom
3595 confidence Limits for slope 0.0706 to 0.3862
95 confidence Limits for intercept -4.34 to 17.85
36- Testing the positive slope
-
The test statistic is
37Reject
df 11 2 9
A one tailed test
38we reject
and conclude
39Confidence Limits for Points on the Regression
Line
- The intercept a is a specific point on the
regression line. - It is the y coordinate of the point on the
regression line when x 0. - It is the predicted value of y when x 0.
- We may also be interested in other points on the
regression line. e.g. when x x0 - In this case the y coordinate of the point on
the regression line when x x0 is a b x0
40x0
41- (1- a)100 Confidence Limits for a b x0
ta/2 is the a/2 critical value for the
t-distribution with n - 2 degrees of freedom
42Prediction Limits for new values of the Dependent
variable y
- An important application of the regression line
is prediction. - Knowing the value of x (x0) what is the value of
y? - The predicted value of y when x x0 is
- This in turn can be estimated by.
43- The predictor
- Gives only a single value for y.
- A more appropriate piece of information would be
a range of values. - A range of values that has a fixed probability of
capturing the value for y. - A (1- a)100 prediction interval for y.
44- (1- a)100 Prediction Limits for y when x x0
ta/2 is the a/2 critical value for the
t-distribution with n - 2 degrees of freedom
45Example
- In this example we are studying building fires in
a city and interested in the relationship
between
- X the distance of the closest fire hall and
the building that puts out the alarm
and
- Y cost of the damage (1000)
The data was collected on n 15 fires.
46The Data
47Scatter Plot
48Computations
49Computations Continued
50Computations Continued
51Computations Continued
52- 95 Confidence Limits for slope b
4.07 to 5.77
t.025 2.160 critical value for the
t-distribution with 13 degrees of freedom
53- 95 Confidence Limits for intercept a
7.21 to 13.35
t.025 2.160 critical value for the
t-distribution with 13 degrees of freedom
54Least Squares Line
55- (1- a)100 Confidence Limits for a b x0
ta/2 is the a/2 critical value for the
t-distribution with n - 2 degrees of freedom
5695 Confidence Limits for a b x0
5795 Confidence Limits for a b x0
58- (1- a)100 Prediction Limits for y when x x0
ta/2 is the a/2 critical value for the
t-distribution with n - 2 degrees of freedom
5995 Prediction Limits for y when x x0
6095 Prediction Limits for y when x x0
61Linear RegressionSummary
- Hypothesis testing and Estimation
62- (1 a)100 Confidence Limits for slope b
ta/2 critical value for the t-distribution with n
2 degrees of freedom
63The test statistic is
- has a t distribution with df n 2 if H0 is
true.
64- (1 a)100 Confidence Limits for intercept a
ta/2 critical value for the t-distribution with n
2 degrees of freedom
65The test statistic is
- has a t distribution with df n 2 if H0 is
true.
66- (1- a)100 Confidence Limits for a b x0
ta/2 is the a/2 critical value for the
t-distribution with n - 2 degrees of freedom
67- (1- a)100 Prediction Limits for y when x x0
ta/2 is the a/2 critical value for the
t-distribution with n - 2 degrees of freedom
68Correlation
69Definition
The statistic
is called Pearsons correlation coefficient
70Properties
- -1 r 1, r 1, r2 1
- r 1 (r 1 or -1) if the points
- (x1, y1), (x2, y2), , (xn, yn) lie along a
straight line. (positive slope for 1, negative
slope for -1)
71The test for independence (zero correlation)
H0 X and Y are independent HA X and Y are
correlated
The test statistic
The Critical region
Reject H0 if t gt ta/2 (df n 2)
This is a two-tailed critical region, the
critical region could also be one-tailed
72Example
- In this example we are studying building fires in
a city and interested in the relationship
between
- X the distance of the closest fire hall and
the building that puts out the alarm
and
- Y cost of the damage (1000)
The data was collected on n 15 fires.
73The Data
74Scatter Plot
75Computations
76Computations Continued
77Computations Continued
78The correlation coefficient
The test for independence (zero correlation)
The test statistic
We reject H0 independence, if t gt t0.025
2.160
H0 independence, is rejected
79Relationship between Regression and Correlation
80Recall
Also
since
Thus the slope of the least squares line is
simply the ratio of the standard deviations the
correlation coefficient
81The test for independence (zero correlation)
H0 X and Y are independent HA X and Y are
correlated
Uses the test statistic
Note
and
82The two tests
- The test for independence (zero correlation)
H0 X and Y are independent HA X and Y are
correlated
- The test for zero slope
H0 b 0. HA b ? 0
are equivalent
83- the test statistic for independence
84Regression (in general)
85- In many experiments we would have collected data
on a single variable Y (the dependent variable )
and on p (say) other variables X1, X2, X3, ... ,
Xp (the independent variables). -
- One is interested in determining a model that
describes the relationship between Y (the
response (dependent) variable) and X1, X2, , Xp
(the predictor (independent) variables. - This model can be used for
- Prediction
- Controlling Y by manipulating X1, X2, , Xp
-
86-
- The Model
- is an equation of the form
- Y f(X1, X2,... ,Xp q1, q2, ... , qq) e
- where q1, q2, ... , qq are unknown parameters of
the function f and e is a random disturbance
(usually assumed to have a normal distribution
with mean 0 and standard deviation s).
87- Examples
- Y Blood Pressure, X age
- The model
- Y a bX e,thus q1 a and q2 b.
- This model is called
- the simple Linear Regression Model
88- Y average of five best times for running the
100m, X the year - The model
- Y a e-bX g e, thus q1 a, q2 b and q2
g. - This model is called
- the exponential Regression Model
Y a e-bX g
89- Y gas mileage ( mpg) of a car brand
- X1 engine size
- X2 horsepower
- X3 weight
- The model
- Y b0 b1 X1 b2 X2 b3 X3 e.
- This model is called
- the Multiple Linear Regression Model
90The Multiple Linear Regression Model
91- In Multiple Linear Regression we assume the
following model -
- Y b0 b1 X1 b2 X2 ... bp Xp e
-
- This model is called the Multiple Linear
Regression Model. - Again are unknown parameters of the model and
where b0, b1, b2, ... , bp are unknown
parameters and e is a random disturbance assumed
to have a normal distribution with mean 0 and
standard deviation s.
92The importance of the Linear model
- 1. It is the simplest form of a model in
which each dependent variable has some effect on
the independent variable Y. - When fitting models to data one tries to find the
simplest form of a model that still adequately
describes the relationship between the dependent
variable and the independent variables. - The linear model is sometimes the first model to
be fitted and only abandoned if it turns out to
be inadequate.
93- In many instance a linear model is the most
appropriate model to describe the dependence
relationship between the dependent variable and
the independent variables. - This will be true if the dependent variable
increases at a constant rate as any or the
independent variables is increased while holding
the other independent variables constant.
94- 3. Many non-Linear models can be Linearized
(put into the form of a Linear model by
appropriately transformation the dependent
variables and/or any or all of the independent
variables.) - This important fact ensures the wide utility of
the Linear model. (i.e. the fact the many
non-linear models are linearizable.)
95An Example
- The following data comes from an experiment that
was interested in investigating the source from
which corn plants in various soils obtain their
phosphorous. - The concentration of inorganic phosphorous (X1)
and the concentration of organic phosphorous (X2)
was measured in the soil of n 18 test plots. - In addition the phosphorous content (Y) of corn
grown in the soil was also measured. The data is
displayed below
96 97Equation Y 56.2510241 1.78977412 X1
0.08664925 X2
98(No Transcript)
99The Multiple Linear Regression Model
100- In Multiple Linear Regression we assume the
following model -
- Y b0 b1 X1 b2 X2 ... bp Xp e
-
- This model is called the Multiple Linear
Regression Model. - Again are unknown parameters of the model and
where b0, b1, b2, ... , bp are unknown
parameters and e is a random disturbance assumed
to have a normal distribution with mean 0 and
standard deviation s.
101Summary of the Statistics used in Multiple
Regression
102- The Least Squares Estimates
- the values that minimize
103- The Analysis of Variance Table Entries
- a) Adjusted Total Sum of Squares (SSTotal)
-
- b) Residual Sum of Squares (SSError)
-
- c) Regression Sum of Squares (SSReg)
-
- Note
-
- i.e. SSTotal SSReg SSError
-
104The Analysis of Variance Table
- Source Sum of Squares d.f. Mean Square F
-
- Regression SSReg p SSReg/p MSReg MSReg/s2
- Error SSError n-p-1 SSError/(n-p-1) MSError
s2 -
- Total SSTotal n-1
105Uses
- 1. To estimate s2 (the error variance).
- - Use s2 MSError to estimate s2.
- To test the Hypothesis
- H0 b1 b2 ... bp 0.
- Use the test statistic
-
- Reject H0 if F gt Fa(p,n-p-1).
106- 3. To compute other statistics that are useful in
describing the relationship between Y (the
dependent variable) and X1, X2, ... ,Xp (the
independent variables). - a) R2 the coefficient of determination
- SSReg/SSTotal
-
- the proportion of variance in Y explained by
- X1, X2, ... ,Xp
- 1 - R2 the proportion of variance in Y
- that is left unexplained by X1, X2, ... , Xp
- SSError/SSTotal.
107- b) Ra2 "R2 adjusted" for degrees of freedom.
- 1 -the proportion of variance in Y that is
left - unexplained by X1, X2,... , Xp adjusted
for d.f.
108- c) R ÖR2 the Multiple correlation
coefficient of Y with X1, X2, ... ,Xp -
- the maximum correlation between Y and a
linear combination of X1, X2, ... ,Xp - Comment The statistics F, R2, Ra2 and R are
equivalent statistics.
109Using Statistical Packages
- To perform Multiple Regression
110Using SPSS
Note The use of another statistical package such
as Minitab is similar to using SPSS
111After starting the SSPS program the following
dialogue box appears
112If you select Opening an existing file and press
OK the following dialogue box appears
113The following dialogue box appears
114If the variable names are in the file ask it to
read the names. If you do not specify the Range
the program will identify the Range
Once you click OK, two windows will appear
115One that will contain the output
116The other containing the data
117To perform any statistical Analysis select the
Analyze menu
118Then select Regression and Linear.
119The following Regression dialogue box appears
120Select the Dependent variable Y.
121Select the Independent variables X1, X2, etc.
122If you select the Method - Enter.
123- All variables will be put into the equation.
There are also several other methods that can be
used
- Forward selection
- Backward Elimination
- Stepwise Regression
124(No Transcript)
125- This method starts with no variables in the
equation - Carries out statistical tests on variables not in
the equation to see which have a significant
effect on the dependent variable. - Adds the most significant.
- Continues until all variables not in the equation
have no significant effect on the dependent
variable.
126- This method starts with all variables in the
equation - Carries out statistical tests on variables in the
equation to see which have no significant effect
on the dependent variable. - Deletes the least significant.
- Continues until all variables in the equation
have a significant effect on the dependent
variable.
127- Stepwise Regression (uses both forward and
backward techniques)
- This method starts with no variables in the
equation - Carries out statistical tests on variables not in
the equation to see which have a significant
effect on the dependent variable. - It then adds the most significant.
- After a variable is added it checks to see if any
variables added earlier can now be deleted. - Continues until all variables not in the equation
have no significant effect on the dependent
variable.
128- All of these methods are procedures for
attempting to find the best equation
The best equation is the equation that is the
simplest (not containing variables that are not
important) yet adequate (containing variables
that are important)
129Once the dependent variable, the independent
variables and the Method have been selected if
you press OK, the Analysis will be performed.
130The output will contain the following table
R2 and R2 adjusted measures the proportion of
variance in Y that is explained by X1, X2, X3,
etc (67.6 and 67.3)
R is the Multiple correlation coefficient (the
maximum correlation between Y and a linear
combination of X1, X2, X3, etc)
131The next table is the Analysis of Variance Table
The F test is testing if the regression
coefficients of the predictor variables are all
zero. Namely none of the independent variables
X1, X2, X3, etc have any effect on Y
132The final table in the output
Gives the estimates of the regression
coefficients, there standard error and the t test
for testing if they are zeroNote Engine size
has no significant effect on Mileage
133The estimated equation from the table below
Is
134Note the equation is
Mileage decreases with
- With increases in Engine Size (not significant, p
0.432)With increases in Horsepower
(significant, p 0.000)With increases in Weight
(significant, p 0.000)
135The Multiple Linear Regression ModelSummary
136- In many experiments we would have collected data
on a single variable Y (the dependent variable )
and on p (say) other variables X1, X2, X3, ... ,
Xp (the independent variables). -
- One is interested in determining a model that
describes the relationship between Y (the
response (dependent) variable) and X1, X2, , Xp
(the predictor (independent) variables. - This model can be used for
- Prediction
- Controlling Y by manipulating X1, X2, , Xp
-
137- In Multiple Linear Regression we assume the
following model -
- Y b0 b1 X1 b2 X2 ... bp Xp e
-
- This model is called the Multiple Linear
Regression Model. - Again are unknown parameters of the model and
where b0, b1, b2, ... , bp are unknown
parameters and e is a random disturbance assumed
to have a normal distribution with mean 0 and
standard deviation s.
138The Statistics in Multiple Regression
139- The Least Squares Estimates
- the values that minimize
140- The Analysis of Variance Table Entries
- a) Adjusted Total Sum of Squares (SSTotal)
-
- b) Residual Sum of Squares (SSError)
-
- c) Regression Sum of Squares (SSReg)
-
- Note
-
- i.e. SSTotal SSReg SSError
-
141The Analysis of Variance Table
- Source Sum of Squares d.f. Mean Square F
-
- Regression SSReg p SSReg/p MSReg MSReg/s2
- Error SSError n-p-1 SSError/(n-p-1) MSError
s2 -
- Total SSTotal n-1
142- Important Summary Statistics
- a) R2 the coefficient of determination
- SSReg/SSTotal
-
- the proportion of variance in Y explained by
- X1, X2, ... ,Xp
- 1 - R2 the proportion of variance in Y
- that is left unexplained by X1, X2, ... , Xp
- SSError/SSTotal.
143- b) Ra2 "R2 adjusted" for degrees of freedom.
- 1 -the proportion of variance in Y that is
left - unexplained by X1, X2,... , Xp adjusted
for d.f.
144- c) R ÖR2 the Multiple correlation
coefficient of Y with X1, X2, ... ,Xp -
- the maximum correlation between Y and a
linear combination of X1, X2, ... ,Xp
145Example
- In this example we are interested in how
- Y mileage (mpg)
- depends on
- X1 engine size
- X2 vehicle weight
- X3 engine horse power
146The output from SPSS
R2 and R2 adjusted measures the proportion of
variance in Y that is explained by X1, X2, X3,
etc (67.6 and 67.3)
R is the Multiple correlation coefficient (the
maximum correlation between Y and a linear
combination of X1, X2, X3, etc)
147The next table is the Analysis of Variance Table
The F test is testing if the regression
coefficients of the predictor variables are all
zero. Namely none of the independent variables
X1, X2, X3, etc have any effect on Y
148The final table in the output
Gives the estimates of the regression
coefficients, there standard error and the t test
for testing if they are zeroNote Engine size
has no significant effect on Mileage
149The estimated equation from the table below
Is
150Note the equation is
Mileage decreases with
- With increases in Engine Size (not significant, p
0.432)With increases in Horsepower
(significant, p 0.000)With increases in Weight
(significant, p 0.000)
151Logistic regression
152- Recall the simple linear regression model
- y b0 b1x e
where we are trying to predict a continuous
dependent variable y from a continuous
independent variable x.
This model can be extended to Multiple linear
regression model y b0 b1x1 b2x2
bpxp e
Here we are trying to predict a continuous
dependent variable y from a several continuous
dependent variables x1 , x2 , , xp .
153Now suppose the dependent variable y is binary.
It takes on two values Success (1) or
Failure (0)
We are interested in predicting a y from a
continuous dependent variable x.
This is the situation in which Logistic
Regression is used
154Example
- We are interested how the success (y) of a new
antibiotic cream is curing acne problems and
how it depends on the amount (x) that is applied
daily. - The values of y are 1 (Success) or 0 (Failure).
- The values of x range over a continuum
155The logisitic Regression Model
- Let p denote Py 1 PSuccess.
- This quantity will increase with the value of x.
is called the odds ratio
The ratio
This quantity will also increase with the value
of x, ranging from zero to infinity.
The quantity
is called the log odds ratio
156Example odds ratio, log odds ratio
- Suppose a die is rolled
- Success roll a six, p 1/6
The odds ratio
The log odds ratio
157The logisitic Regression Model
Assumes the log odds ratio is linearly related to
x.
i. e.
In terms of the odds ratio
158The logisitic Regression Model
Solving for p in terms x.
or
159Interpretation of the parameter b0 (determines
the intercept)
p
x
160Interpretation of the parameter b1 (determines
when p is 0.50 (along with b0))
p
when
x
161Also
when
is the rate of increase in p with respect to x
when p 0.50
162Interpretation of the parameter b1 (determines
slope when p is 0.50 )
p
x
163The data
- The data will for each case consist of
- a value for x, the continuous independent
variable - a value for y (1 or 0) (Success or Failure)
Total of n 250 cases
164(No Transcript)
165Estimation of the parameters
- The parameters are estimated by Maximum
Likelihood estimation and require a statistical
package such as SPSS
166Using SPSS to perform Logistic regression
167- Choose from the menu
- Analyze -gt Regression -gt Binary Logistic
168- The following dialogue box appears
Select the dependent variable (y) and the
independent variable (x) (covariate). Press OK.
169The Estimates and their S.E.
170The parameter Estimates
171Interpretation of the parameter b0 (determines
the intercept)
Interpretation of the parameter b1 (determines
when p is 0.50 (along with b0))
172Another interpretation of the parameter b1
is the rate of increase in p with respect to x
when p 0.50
173The Logistic Regression Model
The dependent variable y is binary. It takes on
two values Success (1) or Failure (0)
We are interested in predicting a y from a
continuous dependent variable x.
174The logisitic Regression Model
- Let p denote Py 1 PSuccess.
- This quantity will increase with the value of x.
is called the odds ratio
The ratio
This quantity will also increase with the value
of x, ranging from zero to infinity.
The quantity
is called the log odds ratio
175The logisitic Regression Model
Assumes the log odds ratio is linearly related to
x.
i. e.
In terms of the odds ratio
176The logisitic Regression Model
In terms of p
177The graph of p vs x
p
x
178The Multiple Logistic Regression model
179- Here we attempt to predict the outcome of a
binary response variable Y from several
independent variables X1, X2 , etc
180Multiple Logistic Regression an example
- In this example we are interested in determining
the risk of infants (who were born prematurely)
of developing BPD (bronchopulmonary dysplasia) - More specifically we are interested in developing
a predictive model which will determine the
probability of developing BPD from - X1 gestational Age and X2 Birthweight
181- For n 223 infants in prenatal ward the
following measurements were determined
- X1 gestational Age (weeks),
- X2 Birth weight (grams) and
- Y presence of BPD
182The data
183The results
184Graph Showing Risk of BPD vs GA and BrthWt
185Non-Parametric Statistics