ISYE 6501 MIDTERM 1 FINAL PAPER COMPLETE
QUESTIONS AND ANSWERS
◉ What types of models prefer standardized data.
Answer: PCA, clustering
◉ What is a possible effect of using unscaled data in a classifier
model.
Answer: introduce bias into the distance metric. Changes in features
with larger ranges will have a disproportionate impact
◉ When should you use KNN over SVM?.
Answer: >2 classes
◉ What is the KNN model sensitive to?.
Answer: noisy predictors and near-0 variance features
◉ KNN: A low k-value may _______.
Answer: overfit
◉ KNN: A high k-value may _______.
Answer: underfit
,◉ KNN: If there are a large number of (possibly irrelevant) features,
a _______ k-value tends to be better..
Answer: larger
◉ KNN: what is the benefit of using an odd vs. even number for k?.
Answer: Odd number of k avoids ties
◉ How is feature importance determined for KNN?.
Answer: features with smallest distance measure
◉ What values do you need to standardize data?.
Answer: mean, standard deviation
◉ What do you do when you want a HARD classifier to take the cost
of misclassification (a more costly error) into account?.
Answer: alter the intercept a0
2/3(a0+1) 1/3(a0-1)
◉ What do you do when you want a SOFT classifier to take the cost
of misclassification (a more costly error) into account?.
Answer: multiply the first half of the equation (error minimization)
by m(j)
, ◉ What is the ∞ norm distance?.
Answer: set p = ∞ in norm distance equation. the sum will equal the
max |xn-yn|.
◉ What does p equal for manhattan distance?.
Answer: 1
◉ What does p equal for euclidean distance?.
Answer: 2
◉ What is x(ij) in the clustering algorithm?.
Answer: attribute j of data point i
◉ What is z(jk) in the clustering algorithm?.
Answer: coordinate j of cluster center k
◉ What is y(ik) in the clustering algorithm?.
Answer: = 1 if data point i is in cluster k
=0 if not
◉ Describe what the K-means algorithm does step-by-step..
QUESTIONS AND ANSWERS
◉ What types of models prefer standardized data.
Answer: PCA, clustering
◉ What is a possible effect of using unscaled data in a classifier
model.
Answer: introduce bias into the distance metric. Changes in features
with larger ranges will have a disproportionate impact
◉ When should you use KNN over SVM?.
Answer: >2 classes
◉ What is the KNN model sensitive to?.
Answer: noisy predictors and near-0 variance features
◉ KNN: A low k-value may _______.
Answer: overfit
◉ KNN: A high k-value may _______.
Answer: underfit
,◉ KNN: If there are a large number of (possibly irrelevant) features,
a _______ k-value tends to be better..
Answer: larger
◉ KNN: what is the benefit of using an odd vs. even number for k?.
Answer: Odd number of k avoids ties
◉ How is feature importance determined for KNN?.
Answer: features with smallest distance measure
◉ What values do you need to standardize data?.
Answer: mean, standard deviation
◉ What do you do when you want a HARD classifier to take the cost
of misclassification (a more costly error) into account?.
Answer: alter the intercept a0
2/3(a0+1) 1/3(a0-1)
◉ What do you do when you want a SOFT classifier to take the cost
of misclassification (a more costly error) into account?.
Answer: multiply the first half of the equation (error minimization)
by m(j)
, ◉ What is the ∞ norm distance?.
Answer: set p = ∞ in norm distance equation. the sum will equal the
max |xn-yn|.
◉ What does p equal for manhattan distance?.
Answer: 1
◉ What does p equal for euclidean distance?.
Answer: 2
◉ What is x(ij) in the clustering algorithm?.
Answer: attribute j of data point i
◉ What is z(jk) in the clustering algorithm?.
Answer: coordinate j of cluster center k
◉ What is y(ik) in the clustering algorithm?.
Answer: = 1 if data point i is in cluster k
=0 if not
◉ Describe what the K-means algorithm does step-by-step..
