Social Sciences Economics Econometrics
ISYE 6501 -Exam 2 Wks 8 - 12
Save
Students also studied
Flashcard sets Study guides Practice tests
ISYE 6501 - Midterm 2 GATECH Midterm 2 Analytics Stats
160 terms 51 terms Teacher 53 terms
oanise Preview micahlowens9 Preview mutulani
Building simpler models with A. Overfitting
fewer factors helps avoid which D. Difficulty of interpretation
problems?
A. Overfitting
B. Low prediction quality
C. Bias in the most important
factors
D. Difficulty in interpretation
Two main reasons to limit # of 1. Overfitting
factors in a model. 2. Simplicity
When is overfitting likely to When the number of factors is close to the number of data
happen? points.
How does using a # of factors The model too closely fits the random efffects. It fits that data set
that is close to the number of well, but fails to predict well on a new data set.
data points cause overfitting?
1. Less data is required
Three reasons simple models
2. Less chance of including insignificant factors
are better than complex models.
3. Easier to interpret
Which of these is a key A. If the data isn't scaled, the coefficients can have artificially
difference between stepwise different orders of magnitude, which means they'll have
regression and lasso regression? unbalanced effects on the lasso constraint.
A. Lasso regression requires the
data to first be scaled
B. Stepwise regression gives
many models to choose from,
while lasso gives just one.
1. Forward Selection
Name three greedy approaches
2. Backward Elimination
to variable selection.
3. Stepwise Regression
Name two global approaches to 1. Lasso
variable selection. 2. Elastic Net
, At each step, the model does one thing that looks best without
What does a greedy approach
taking future options into consideration. A more classical
to variable selection mean?
approach.
How does forward selection Forward selection starts with zero factors and backward
differ from backward elimination starts with all factors.
elimination?
A combination of forward selection and backward elimination. At
What is stepwise regression? each step, a variable is considered for addition or elimination
based on some prespecified criterion.
Name two approaches to use 1. Start with all factors
with stepwise regression. 2. Start with no factors
What restriction does the lasso The sum of the coefficients can't be too large. "t"
approach add?
It uses that on the most important coefficients and the others will
How does lasso use the t value?
be zero so those factors won't be part of the model.
What do you need to with the Scale the data.
data whenever you're
constraining the sum of
coefficients?
The value of t in the lasso 1. The number of variables you want.
approach depends on what two 2. The quality of the model as you allow more variables
things?
What is the best approach to Try lasso with different values of t and pick the value that has the
find the best value of t with best tradeoff between number of variables and quality of the
lasso? model.
What constraint does Elastic Net Elastic Net constrains the absolute value of the coefficients and
add? their squares.
Name two things similar about 1. You have to scale the data for both.
Lasso and Elastic Net. 2. You have to pick the best value of t for both.
Do you have to scale the data Yes
for Elastic Net?
What change can you make to Remove the absolute value term.
elastic net to get a model called
ridge regression?
Does ridge regression do No.
variable selection?
What is the advantage of ridge It can lead to better predictive models.
regression?
When two predictors are highly B. Ridge regression will choose smaller (in an absolute sense)
correlated, which of the non-zero coefficients for both models. By nature, it may
following statements is true? underestimate the effect of the factors.
A. Lasso regression will usually
have non-zero coefficients for
both predictors.
B. Ridge regression will usually
have non-zero coefficients for
both predictors.
ISYE 6501 -Exam 2 Wks 8 - 12
Save
Students also studied
Flashcard sets Study guides Practice tests
ISYE 6501 - Midterm 2 GATECH Midterm 2 Analytics Stats
160 terms 51 terms Teacher 53 terms
oanise Preview micahlowens9 Preview mutulani
Building simpler models with A. Overfitting
fewer factors helps avoid which D. Difficulty of interpretation
problems?
A. Overfitting
B. Low prediction quality
C. Bias in the most important
factors
D. Difficulty in interpretation
Two main reasons to limit # of 1. Overfitting
factors in a model. 2. Simplicity
When is overfitting likely to When the number of factors is close to the number of data
happen? points.
How does using a # of factors The model too closely fits the random efffects. It fits that data set
that is close to the number of well, but fails to predict well on a new data set.
data points cause overfitting?
1. Less data is required
Three reasons simple models
2. Less chance of including insignificant factors
are better than complex models.
3. Easier to interpret
Which of these is a key A. If the data isn't scaled, the coefficients can have artificially
difference between stepwise different orders of magnitude, which means they'll have
regression and lasso regression? unbalanced effects on the lasso constraint.
A. Lasso regression requires the
data to first be scaled
B. Stepwise regression gives
many models to choose from,
while lasso gives just one.
1. Forward Selection
Name three greedy approaches
2. Backward Elimination
to variable selection.
3. Stepwise Regression
Name two global approaches to 1. Lasso
variable selection. 2. Elastic Net
, At each step, the model does one thing that looks best without
What does a greedy approach
taking future options into consideration. A more classical
to variable selection mean?
approach.
How does forward selection Forward selection starts with zero factors and backward
differ from backward elimination starts with all factors.
elimination?
A combination of forward selection and backward elimination. At
What is stepwise regression? each step, a variable is considered for addition or elimination
based on some prespecified criterion.
Name two approaches to use 1. Start with all factors
with stepwise regression. 2. Start with no factors
What restriction does the lasso The sum of the coefficients can't be too large. "t"
approach add?
It uses that on the most important coefficients and the others will
How does lasso use the t value?
be zero so those factors won't be part of the model.
What do you need to with the Scale the data.
data whenever you're
constraining the sum of
coefficients?
The value of t in the lasso 1. The number of variables you want.
approach depends on what two 2. The quality of the model as you allow more variables
things?
What is the best approach to Try lasso with different values of t and pick the value that has the
find the best value of t with best tradeoff between number of variables and quality of the
lasso? model.
What constraint does Elastic Net Elastic Net constrains the absolute value of the coefficients and
add? their squares.
Name two things similar about 1. You have to scale the data for both.
Lasso and Elastic Net. 2. You have to pick the best value of t for both.
Do you have to scale the data Yes
for Elastic Net?
What change can you make to Remove the absolute value term.
elastic net to get a model called
ridge regression?
Does ridge regression do No.
variable selection?
What is the advantage of ridge It can lead to better predictive models.
regression?
When two predictors are highly B. Ridge regression will choose smaller (in an absolute sense)
correlated, which of the non-zero coefficients for both models. By nature, it may
following statements is true? underestimate the effect of the factors.
A. Lasso regression will usually
have non-zero coefficients for
both predictors.
B. Ridge regression will usually
have non-zero coefficients for
both predictors.
