ISYE6414 FINAL EXAM / ISYE6414 FINAL EXAM REAL EXAM
ISYE 6414 Final Exam
QUESTIONS AND 100% CORRECT ANSWERS PLUS RATIONALES/
Study online at https://quizlet.com/_hcbj9d
GRADED A
1. In Logistic Regression, the relation- True - The relationship that links the predictors is
ship between the probability of suc- highly non-linear.
cess and the predicting variables is
non-linear.
2. In Logistic Regression, the error terms False - In logistic regression, there are no error
follow a normal distribution. terms.
3. The logit function is the log of the ra- True - the logit function is also known as the
tio of the probability of success to the log-odds function, which is the ln(P/1-p).
probability of failure and is also known
as the log-odds function.
4. The number of parameters that need False - As there is no error term in logistic re-
to be estimated in a logistic regression gression, there is no additional parameter for the
model with 6 predicting variables and variance of the error terms.
an intercept is the same as the num-
ber of parameters that need to be esti-
mated in a standard linear regression
model with an intercept and same pre-
dicting variables.
5. The log-likelihood function is a linear False - log-likelihood is a non-linear function, and a
function with a closed form solution. numerical algorithm is needed in order to maximize
it.
6. In Logistic Regression, the estimat- False - We interpret logistic regression coefficients
ed value for a regression coefficient with respect to the odds of success.
B represents the estimated expected
change in the response variable asso-
ciated with a one unit increase in the
predicting variable, holding all else
fixed.
, ISYE 6414 Final Exam
Study online at https://quizlet.com/_hcbj9d
7. Under logistic regression, the sam- False - The coefficient estimator follows an approx-
pling distribution used for a coeffi- imate normal distribution.
cient estimator is a chi-square distrib-
ution when the sample size is large.
8. When testing a subset of coefficients, False - when testing a subset of coefficients, de-
deviance follows a chi-square distribu- viance follows a chi-square distribution with q de-
tion with q degrees of freedom, where grees of freedom, where q is the number of regres-
q is the number of regression coeffi- sion coefficients discarded from the full model to
cients in the reduced model. get the reduced model.
9. Logistic regression deals with the case True - logistic regression is the generalization of the
where the dependent variable is bina- standard regression model that is used when the
ry and the conditional distribution is response variable y is binary or binomial.
binomial.
10. It is good practice to perform a good- False - The residuals can only be defined for logistic
ness-of-fit test on logistic regression regression with replications.
models without replications.
11. In Logistic regression, if the p-value of False - for logistic regression, if the p-value of the
the deviance test for GOF is smaller deviance test for GOD is large, then the model is a
than the significance level alpha, then good fit.
is is plausible that the model is a good
fit.
12. If a logistic regression model provides False - GOF is no guarantee for good prediction
accurate classification, then we can and vice-versa.
conclude that it is a good fir for the
data.
13.
ISYE 6414 Final Exam
QUESTIONS AND 100% CORRECT ANSWERS PLUS RATIONALES/
Study online at https://quizlet.com/_hcbj9d
GRADED A
1. In Logistic Regression, the relation- True - The relationship that links the predictors is
ship between the probability of suc- highly non-linear.
cess and the predicting variables is
non-linear.
2. In Logistic Regression, the error terms False - In logistic regression, there are no error
follow a normal distribution. terms.
3. The logit function is the log of the ra- True - the logit function is also known as the
tio of the probability of success to the log-odds function, which is the ln(P/1-p).
probability of failure and is also known
as the log-odds function.
4. The number of parameters that need False - As there is no error term in logistic re-
to be estimated in a logistic regression gression, there is no additional parameter for the
model with 6 predicting variables and variance of the error terms.
an intercept is the same as the num-
ber of parameters that need to be esti-
mated in a standard linear regression
model with an intercept and same pre-
dicting variables.
5. The log-likelihood function is a linear False - log-likelihood is a non-linear function, and a
function with a closed form solution. numerical algorithm is needed in order to maximize
it.
6. In Logistic Regression, the estimat- False - We interpret logistic regression coefficients
ed value for a regression coefficient with respect to the odds of success.
B represents the estimated expected
change in the response variable asso-
ciated with a one unit increase in the
predicting variable, holding all else
fixed.
, ISYE 6414 Final Exam
Study online at https://quizlet.com/_hcbj9d
7. Under logistic regression, the sam- False - The coefficient estimator follows an approx-
pling distribution used for a coeffi- imate normal distribution.
cient estimator is a chi-square distrib-
ution when the sample size is large.
8. When testing a subset of coefficients, False - when testing a subset of coefficients, de-
deviance follows a chi-square distribu- viance follows a chi-square distribution with q de-
tion with q degrees of freedom, where grees of freedom, where q is the number of regres-
q is the number of regression coeffi- sion coefficients discarded from the full model to
cients in the reduced model. get the reduced model.
9. Logistic regression deals with the case True - logistic regression is the generalization of the
where the dependent variable is bina- standard regression model that is used when the
ry and the conditional distribution is response variable y is binary or binomial.
binomial.
10. It is good practice to perform a good- False - The residuals can only be defined for logistic
ness-of-fit test on logistic regression regression with replications.
models without replications.
11. In Logistic regression, if the p-value of False - for logistic regression, if the p-value of the
the deviance test for GOF is smaller deviance test for GOD is large, then the model is a
than the significance level alpha, then good fit.
is is plausible that the model is a good
fit.
12. If a logistic regression model provides False - GOF is no guarantee for good prediction
accurate classification, then we can and vice-versa.
conclude that it is a good fir for the
data.
13.
