ISLR EXAM QUESTIONS WITH
CORRECT ANSWERS
to minimize test error, need to achieve ______ and _____ - ANSWER-low variance;
low bias
variance definition - ANSWER-amount f carat (estimate) would change if estimated
on a different training set
bias definition - ANSWER-error introduced by approximating a real-life problem (e.g.
linear regression)
As flexibility increases, bias _________ __________ than variance __________ -
ANSWER-decreases faster; increases
one of most common methods of quantifying accuracy (in classification f carat
estimates) - ANSWER-error rate
indicator variable - ANSWER-binary explanatory variable (with two levels)
training error - ANSWER-computed using data used to training classifier
test error - ANSWER-Ave(I(y0 != ycarat0))
Conditional probability - ANSWER-Given x = x0, predict that Y=class j
Bayes classifier - ANSWER-P(Y=j | X=x0) > .5 then class 1; otherwise class 2
Bayes decision boundary - ANSWER-Boundary that marks where observations get
classified into class1 vs. class2
Bayes error rate - ANSWER-
KNN classifies x0 to ________________ - ANSWER-class j with highest probability
using p(k = n neighbors)
p-value - ANSWER-measure for whether to reject null hypothesis or not
metrics for quantifying the extent of how well model fits the data - ANSWER-R^2 and
RSE
RSE is the ________ of e (error term) - ANSWER-standard deviation
RSE measures average amount __________ - ANSWER-average amount the
response will deviate from the regression line
RSE measures ______ fit in _________ - ANSWER-lack of; the model
CORRECT ANSWERS
to minimize test error, need to achieve ______ and _____ - ANSWER-low variance;
low bias
variance definition - ANSWER-amount f carat (estimate) would change if estimated
on a different training set
bias definition - ANSWER-error introduced by approximating a real-life problem (e.g.
linear regression)
As flexibility increases, bias _________ __________ than variance __________ -
ANSWER-decreases faster; increases
one of most common methods of quantifying accuracy (in classification f carat
estimates) - ANSWER-error rate
indicator variable - ANSWER-binary explanatory variable (with two levels)
training error - ANSWER-computed using data used to training classifier
test error - ANSWER-Ave(I(y0 != ycarat0))
Conditional probability - ANSWER-Given x = x0, predict that Y=class j
Bayes classifier - ANSWER-P(Y=j | X=x0) > .5 then class 1; otherwise class 2
Bayes decision boundary - ANSWER-Boundary that marks where observations get
classified into class1 vs. class2
Bayes error rate - ANSWER-
KNN classifies x0 to ________________ - ANSWER-class j with highest probability
using p(k = n neighbors)
p-value - ANSWER-measure for whether to reject null hypothesis or not
metrics for quantifying the extent of how well model fits the data - ANSWER-R^2 and
RSE
RSE is the ________ of e (error term) - ANSWER-standard deviation
RSE measures average amount __________ - ANSWER-average amount the
response will deviate from the regression line
RSE measures ______ fit in _________ - ANSWER-lack of; the model
