ISYE 6501 -Exam 2 QUESTIONS WITH 100% VERIFIED
SOLUTIONS
What is the difference between the quadratic term in Ridge regression and the absolute
value term in Lasso? - ANSWER Lasso's absolute value term makes some coefficients
equal zero. Ridge regressions quadratic term shrinks the coefficients, but they won't
equal zero.
What does shrinking the coefficient values do and what is the negative trade off? -
ANSWER It helps reduce the variance, but it adds some bias.
What is the benefit of trading off bias for variance? - ANSWER Better predictive models
(prediction error is a function of both bias and variance)
What are the benefits of elastic net? - ANSWER It combines the variable selection
benefits of lasso with the predictive benefits of ridge regression.
It also combines some of the drawbacks of each.
What is a disadvantage to shrinking the coefficients using Ridge regression? -
ANSWER Really good predictors can have their coefficients shrunk and will be
underestimated by the model.
What is a disadvantage to eliminating some variables using lasso? - ANSWER If you
have two highly correlated predictors, lasso might only pick one of them and it might not
be the preferred one (for example: two medical tests. lasso may pick the expensive one
when you'd prefer the inexpensive one to be used in the model)
What is blocking? - ANSWER Blocking is a factor that could create variation. (in the car
example, the type of car is a blocking factor)
If we're testing to see whether red cars sell for higher prices than blue cars, we need to
account for the type and age of the cars in our data set. This is called... - ANSWER
Controlling.
What three things do you need to do A/B testing? - ANSWER 1. Be able to collect data
quickly
2. Data must be representative of the population
3. Amount of data must be small compared to population.
What is A/B testing? - ANSWER A test to see which of two alternatives is better.
, Common in marketing. (testing two banner ads, for example)
What is a full factorial design? - ANSWER Testing the effectiveness of every
combination.
What is a fractional factorial design? - ANSWER Testing a subset of the combinations.
Usually used when there are too many combinations to do a full factorial test.
What is a balanced design for a fractional factorial test? - ANSWER A test design that
tests each choice at least once and each pair of choices shows up the same number of
times.
What is an option to a factorial test and what conditions need to be met? - ANSWER If
the factors are independent, you could test a subset of combinations and use
regression to estimate the effects.
In what stage of analytics do you use factorial designs? - ANSWER Before modeling
and before collecting data.
What does exploration vs exploitation mean? - ANSWER It's the trade-off between
collecting more information and getting immediate value.
What is the multi armed bandit problem? - ANSWER it's a problem in which a fixed
limited set of resources must be allocated between competing (alternative) choices in a
way that maximizes their expected gain, when each choice's properties are only partially
known at the time of allocation, and may become better understood as time passes or
by allocating resources to the choice
Exploitation vs Exploration.
What are some things you can change when doing a multi-armed bandit? - ANSWER -
The number of tests
- The way we update the probabilities (Bayesian or from the observed distribution)
- The way we assign new tests
What are two advantages of multi-armed bandits? - ANSWER 1. Learn faster on the fly
2. They create more value along the way
What is the Bernoulli distribution? - ANSWER A probability distribution of any single
experiment that has two outcomes (yes/no, for example) and the variables are
indpendent.
What happens to the binomial distribution when n gets very large? - ANSWER It
SOLUTIONS
What is the difference between the quadratic term in Ridge regression and the absolute
value term in Lasso? - ANSWER Lasso's absolute value term makes some coefficients
equal zero. Ridge regressions quadratic term shrinks the coefficients, but they won't
equal zero.
What does shrinking the coefficient values do and what is the negative trade off? -
ANSWER It helps reduce the variance, but it adds some bias.
What is the benefit of trading off bias for variance? - ANSWER Better predictive models
(prediction error is a function of both bias and variance)
What are the benefits of elastic net? - ANSWER It combines the variable selection
benefits of lasso with the predictive benefits of ridge regression.
It also combines some of the drawbacks of each.
What is a disadvantage to shrinking the coefficients using Ridge regression? -
ANSWER Really good predictors can have their coefficients shrunk and will be
underestimated by the model.
What is a disadvantage to eliminating some variables using lasso? - ANSWER If you
have two highly correlated predictors, lasso might only pick one of them and it might not
be the preferred one (for example: two medical tests. lasso may pick the expensive one
when you'd prefer the inexpensive one to be used in the model)
What is blocking? - ANSWER Blocking is a factor that could create variation. (in the car
example, the type of car is a blocking factor)
If we're testing to see whether red cars sell for higher prices than blue cars, we need to
account for the type and age of the cars in our data set. This is called... - ANSWER
Controlling.
What three things do you need to do A/B testing? - ANSWER 1. Be able to collect data
quickly
2. Data must be representative of the population
3. Amount of data must be small compared to population.
What is A/B testing? - ANSWER A test to see which of two alternatives is better.
, Common in marketing. (testing two banner ads, for example)
What is a full factorial design? - ANSWER Testing the effectiveness of every
combination.
What is a fractional factorial design? - ANSWER Testing a subset of the combinations.
Usually used when there are too many combinations to do a full factorial test.
What is a balanced design for a fractional factorial test? - ANSWER A test design that
tests each choice at least once and each pair of choices shows up the same number of
times.
What is an option to a factorial test and what conditions need to be met? - ANSWER If
the factors are independent, you could test a subset of combinations and use
regression to estimate the effects.
In what stage of analytics do you use factorial designs? - ANSWER Before modeling
and before collecting data.
What does exploration vs exploitation mean? - ANSWER It's the trade-off between
collecting more information and getting immediate value.
What is the multi armed bandit problem? - ANSWER it's a problem in which a fixed
limited set of resources must be allocated between competing (alternative) choices in a
way that maximizes their expected gain, when each choice's properties are only partially
known at the time of allocation, and may become better understood as time passes or
by allocating resources to the choice
Exploitation vs Exploration.
What are some things you can change when doing a multi-armed bandit? - ANSWER -
The number of tests
- The way we update the probabilities (Bayesian or from the observed distribution)
- The way we assign new tests
What are two advantages of multi-armed bandits? - ANSWER 1. Learn faster on the fly
2. They create more value along the way
What is the Bernoulli distribution? - ANSWER A probability distribution of any single
experiment that has two outcomes (yes/no, for example) and the variables are
indpendent.
What happens to the binomial distribution when n gets very large? - ANSWER It
