MGSC 291 Exam 3 Prep
make date variable be recognized as a date in R with the as.Date() function - ANS-What is the
first step in processing you data for a time series analysis in R?
- make all data factors
- make date variable a factor with the factor() function
-make date variable be recognized as a date in R with the as.Date() function
as the number of days relative to Jan 1, 1970 - ANS-How does R store date variables?
-as factors
-as characters
-as the number of days relative to Jan 1, 1970
seasonal variation - ANS-In a time series, regular variation that is repeated within a year is
called....
-non-stationary
-seasonal variation
-a random walk
diverging series - ANS-Which of the following types of series are useless for modeling?
- random walk
- diverging series
- mean reverting, stationary
stationary, mean reverting - ANS-Which of the following is the most useful type of series for
modeling?
- random walk
- diverging series
- stationary, mean reverting
- ANS-Match the coefficient on the AR(1) term in a regression to the correct series
random walk - ANS-|beta| = 1
- random walk
- mean reverting
-diverging
mean reverting - ANS-|beta| < 1
- random walk
- mean-reverting
-diverging
, diverging - ANS-|beta| > 1
- random walk
- mean reverting
-diverging
use the returns transformation for modeling - ANS-if you have a random walk, what should you
do?
-use the returns transformation for modeling
- incorporate other important trend variables
- scrap the data - this type is useless
- the simple interrelations of the AR(1) term no longer apply
- if you need higher lags, you might have missed an important trend or seasonality -
ANS-(WHICH 2 APPLY) What is true about adding higher than AR(1) lags to a model?
- the simple interrelations of the AR(1) term no longer apply
-this model always works better for prediction than the model with only the AR(1) term
-you almost always need to take a log of the response
- if you need higher lags, you might have missed an important trend or seasonality
- you might have modeled the wrong response
an inelastic good - ANS-A sales price elasticity greater than -1 implies
- an inelastic good
- a good model fit
- a random walk-
-a diverging series
panel data - ANS-When you have multiple stacks of time series, for example a time series of
monthly sales for 185 stores, you have
- a random walk
- a mean reverting time series
- panel data
- no need for fixed effects
-are different from random effects which allow for correlations between the error terms
- are just simply a way to include factor variable into the regression - ANS-(WHICH 2 APPLY)
Fixed effects are:
- effects that fix your residuals to be uncorrelated
-are different from random effects which allow for correlations between the error terms
-are just simply a way to include factor variables into the regression
-make it unnecessary to check model fit since they fix errors in the regression
underestimate of uncertainty - ANS-Dependencies in the data often lead to an
-underestimate of uncertainty
-an overestimate of uncertainty
make date variable be recognized as a date in R with the as.Date() function - ANS-What is the
first step in processing you data for a time series analysis in R?
- make all data factors
- make date variable a factor with the factor() function
-make date variable be recognized as a date in R with the as.Date() function
as the number of days relative to Jan 1, 1970 - ANS-How does R store date variables?
-as factors
-as characters
-as the number of days relative to Jan 1, 1970
seasonal variation - ANS-In a time series, regular variation that is repeated within a year is
called....
-non-stationary
-seasonal variation
-a random walk
diverging series - ANS-Which of the following types of series are useless for modeling?
- random walk
- diverging series
- mean reverting, stationary
stationary, mean reverting - ANS-Which of the following is the most useful type of series for
modeling?
- random walk
- diverging series
- stationary, mean reverting
- ANS-Match the coefficient on the AR(1) term in a regression to the correct series
random walk - ANS-|beta| = 1
- random walk
- mean reverting
-diverging
mean reverting - ANS-|beta| < 1
- random walk
- mean-reverting
-diverging
, diverging - ANS-|beta| > 1
- random walk
- mean reverting
-diverging
use the returns transformation for modeling - ANS-if you have a random walk, what should you
do?
-use the returns transformation for modeling
- incorporate other important trend variables
- scrap the data - this type is useless
- the simple interrelations of the AR(1) term no longer apply
- if you need higher lags, you might have missed an important trend or seasonality -
ANS-(WHICH 2 APPLY) What is true about adding higher than AR(1) lags to a model?
- the simple interrelations of the AR(1) term no longer apply
-this model always works better for prediction than the model with only the AR(1) term
-you almost always need to take a log of the response
- if you need higher lags, you might have missed an important trend or seasonality
- you might have modeled the wrong response
an inelastic good - ANS-A sales price elasticity greater than -1 implies
- an inelastic good
- a good model fit
- a random walk-
-a diverging series
panel data - ANS-When you have multiple stacks of time series, for example a time series of
monthly sales for 185 stores, you have
- a random walk
- a mean reverting time series
- panel data
- no need for fixed effects
-are different from random effects which allow for correlations between the error terms
- are just simply a way to include factor variable into the regression - ANS-(WHICH 2 APPLY)
Fixed effects are:
- effects that fix your residuals to be uncorrelated
-are different from random effects which allow for correlations between the error terms
-are just simply a way to include factor variables into the regression
-make it unnecessary to check model fit since they fix errors in the regression
underestimate of uncertainty - ANS-Dependencies in the data often lead to an
-underestimate of uncertainty
-an overestimate of uncertainty
