Question 1

Maximum likelihood estimation is an advanced statistical procedure which provides point estimates, and it is typically
A) ahead of the logistic regression.
B) a part of the logistic regression.
C) after the logistic regression to ensure statistical accuracy.
D) a simple process, completed to provide comparison estimates.

Accepted Answer

Maximum likelihood estimation is an advanced statistical procedure which provides point estimates and it is typically ahead of the logistic regression.

Question 2

Logistic regression analysis can answer what three major questions about a dataset?
A) Causal analysis, forecasting of an outcome, and trend forecasting
B) Causal analysis, odds of success, and trend forecasting
C) Odds of success, trend forecasting, and the confusion matrix
D) Maximum likelihood, causal analysis, and odds of success

Accepted Answer

Logistic regression analysis can answer causal analysis, forecasting of an outcome, and trend forecasting.

Question 3

Which method will we use to best fit the data in logistic regression?
A) Maximum likelihood
B) Least effect likelihood
C) Least square error
D) Jaccard distance

Accepted Answer

Jaccard distance will best fit the data in logistic regression.

Question 4

An example of a dummy variable is used in categorical data applications and might include
A) assigning the number 1 to Democrats and 2 to Republicans.
B) assigning the number 1 to the first place runner, and 2 to the second place runner.
C) assigning a 0 to students who can run a 10-minute mile, and a number 1 to students who run a 7-minute mile.
D) assigning a 1 to the probability of the customer receiving a $10 gift card and a 0 to the customer receiving a $20 gift card, when 10 of each denomination are being sent to a random sample of customers.

Accepted Answer

Dummy variables assign values to nominal data.

Question 5

One of the very good methods to analyze the performance of logistic regression is AIC, which is similar to R-squared in Linear Regression. Which of the following is true about AIC?
A) Model with minimum AIC value
B) Model with maximum AIC value
C) Both but depends on the situation
D) Both regardless of the situation

Accepted Answer

Model with minimum AIC value.

Question 6

In a neural network, the penalty weight controls the trade-off between ________ and ________.
A) overfitting; underfitting
B) overfitting; not fitting
C) underfitting; fitting
D) fitting; not fitting

Accepted Answer

In a neural network, the penalty weight controls the trade-off between overfitting and underfitting.

Question 7

Deviance, Pearson and Hosmer-Lemeshow are all ________ that can be used in ________.
A) goodness of fit tests; logistic regression
B) predictive analysis tests; logistic regression
C) probability tests; neural network modeling
D) goodness of fit tests; neural network modeling

Accepted Answer

Deviance, Pearson and Hosmer-Lemeshow are all goodness-of-fit tests that can be used in logistic regression.

Question 8

You want to add new features to data to which you applied a logistic regression model and got a training accuracy of X and a testing accuracy of Y. Which options are correct, if you consider the remaining parameters are the same?
A) Training accuracy increases and testing accuracy increases or remains the same.
B) Training accuracy increases or remains the same.
C) Testing accuracy decreases and training accuracy remains the same.
D) Testing accuracy increases or remains the same and training accuracy decreases.

Accepted Answer

X increases in this case, which could (but not necessarily) cause an increase in Y.

Question 9

What is the natural logarithm of the odds called?
A) The logit
B) The odds logarithm
C) The odds logit
D) The natural odds

Accepted Answer

The logit is the natural logarithm of the odds.

Question 10

In a neural network model, the ________ is a nonlinear function of the linear predictor values.
A) response variable
B) null variable
C) predictor variable
D) quantitative response variable

Accepted Answer

In a neural network model, the response variable is a nonlinear function of the linear predictor values.

Question 11

Which type of predictive analytic would be used to determine whether players on a football team, either historical or future, would perform successfully or unsuccessfully in a certain position on the field?
A) Linear discriminate
B) Logistic regression
C) Neural networks
D) Any of the choices could be used.

Accepted Answer

Linear discriminate analysis is used for linear analysis of nominal data.

Question 12

Consider the following model for logistic regression: P (y = 1|x, h) = g(h₀ + h₁x) where g(z) is the logistic function. In the preceding equation the P (y = 1|x; h) , viewed as a function of x, that we can get by changing the parameters h. What would be the range of P in such case? A) (0, 1) B) (0, infinity) C) (−infinity, 0) D) (−infinity, infinity)

Accepted Answer

x itself must be a value between 0 and 1, exclusive, and so must be P.

Question 13

In logistic regression, the difference between a fitted model and the saturated model can be measured with what test?
A) Deviance
B) Pearson
C) Hosmer-Lemeshow
D) All of the choices are correct.

Accepted Answer

Deviance shows the difference between a fitted and saturated model in logistic regression.

Question 14

A logistic curve initially grows exponentially and
A) is theoretical.
B) is a summary of the probabilities.
C) is a consistent curve regardless of the data.
D) reflects the confidence intervals.

Accepted Answer

A logistic curve initially grows exponentially and is theoretical.

Question 15

What are used to avoid overfitting in the neural network model, and are used to minimize a penalized least squares criterion?
A) Parameter estimates
B) Least squares estimates
C) Probability estimates
D) Any of the choice could be used.

Accepted Answer

Parameter estimates are used to avoid overfitting in the neural network model, and are used to minimize a penalized least squares criterion.

Question 16

The odds of an event occurring is: A. The probability that the event will occur divided by the probability that the event will not occur.
B. The probability that the event will not occur divided by the probability that the event will occur.
C. The probability that the event will occur divided by the total of the probability that it will occur, plus the probability of the event not occurring.
A) Option A
B) Option B
C) Option C
D) Either option A or B would create the same outcome.

Accepted Answer

Odds of an event occurring = P(event will occur)/P(event will not occur)

Question 17

A ________ consists of three layers: input, hidden, and output.
A) single-layer perceptron neural network model
B) double-layer perceptron neural network model
C) triple-layer perceptron neural network model
D) neural network model

Accepted Answer

A single-layer perceptron neural network model consists of three layers: input, hidden, and output.

Question 18

If the odds of a student passing a class are .25, what is the probability that the student will pass the class?
A) .20
B) .25
C) .75
D) .80

Accepted Answer

Since the odds are 0.25, 0.25 = 1/4. So, a student will pass 1 class, but fail 4. Thus, the student will pass 1 out of 5 classes, so the probability of passing the class is 1/5 = 0.2

Question 19

Parametric predictive analysis can be used to classify qualitative variables into several groups. The processes might include:
A) logistic regression.
B) simple regression.
C) vertical discriminate analysis.
D) stratification and clustering techniques.

Accepted Answer

Logistic regression is a technique to classify qualitative variables into several groups.

Question 20

What is a major drawback of the neural network modeling?
A) The parameters are typically uninterpretable.
B) The model is not easily utilized with existing data sets.
C) The parameters minimize the least squares criterion.
D) This model only uses a single-layer perceptron.

Accepted Answer

It is virtually impossible to interpret the parameters in neural network modeling.

Quiz 16: Predictive Analytics Ii: Logistic Regression, Discriminate Analysis,

Maximum likelihood estimation is an advanced statistical procedure which provides point estimates, and it is typically

Logistic regression analysis can answer what three major questions about a dataset?

Which method will we use to best fit the data in logistic regression?

An example of a dummy variable is used in categorical data applications and might include

One of the very good methods to analyze the performance of logistic regression is AIC, which is similar to R-squared in Linear Regression. Which of the following is true about AIC?

In a neural network, the penalty weight controls the trade-off between and .

Deviance, Pearson and Hosmer-Lemeshow are all that can be used in .

You want to add new features to data to which you applied a logistic regression model and got a training accuracy of X and a testing accuracy of Y. Which options are correct, if you consider the remaining parameters are the same?

What is the natural logarithm of the odds called?

In a neural network model, the ________ is a nonlinear function of the linear predictor values.

Which type of predictive analytic would be used to determine whether players on a football team, either historical or future, would perform successfully or unsuccessfully in a certain position on the field?

Consider the following model for logistic regression: P (y = 1|x, h) = g(h₀ + h₁x) where g(z) is the logistic function. In the preceding equation the P (y = 1|x; h) , viewed as a function of x, that we can get by changing the parameters h. What would be the range of P in such case?

In logistic regression, the difference between a fitted model and the saturated model can be measured with what test?

A logistic curve initially grows exponentially and

What are used to avoid overfitting in the neural network model, and are used to minimize a penalized least squares criterion?

A ________ consists of three layers: input, hidden, and output.

If the odds of a student passing a class are .25, what is the probability that the student will pass the class?

Parametric predictive analysis can be used to classify qualitative variables into several groups. The processes might include:

What is a major drawback of the neural network modeling?