1|Page
ISYE 6501 Midterm 1 Exam Newest 2025 Complete
100 questions and Correct Detailed Answers (Verified
Answers) |Already Graded A+
A survey of 25 people recorded each person's family size and
type of car. Which of these is a data point?
A. The 14th person's family size and car type
B. The 14th person's family size
C. The car type of each person - ANSWER-A.
observation/record of measurements for a single member of a
population or data set. In the standard tabular format, a row of
data.
The farther the wrongly classified point is from the line...... -
ANSWER-The bigger the mistake we've made
The term including the margin gets larger so the importance of a
large margin out weights avoiding mistakes and classifying
known data samples - ANSWER-As lambda gets larger
Each column is information about what? - ANSWER-An
observation
,2|Page
What is a Support Vector Macine? - ANSWER-Basic Machine
Learning; classifying a binary response based on a set of
attributes and data points.
In SVM, do we scale the data? - ANSWER-Must scale the data!
In SVM, what is C? - ANSWER-C Value or Cost - this is our
tradeoff factor between correctness and margin
What is Box Cox Transformation? - ANSWER-useful for
transforming a response to eliminate heteroscedasticity;
logarithmic transformation; stretches out the smaller range to
enlarge its variability; shrinks the larger range to reduce its
variability. The idea is to find the best value of lambda; first
check the Q-Q plot to see if you need to do it.
What is hteroscedasticity? - ANSWER-Unequal variance; might
cause bias in our data b/c variance at the upper end can make
estimation errors larger and push the model to fit those data
points better
,3|Page
When should detrending be considered? - ANSWER-Response,
Predictors, Factor-based model, regression, SVM, etc. to analyze
time series data
How to detrend - ANSWER-Factor by Factor; i.e.: simple linear
regression
Price = 45,600 +23.2 * Year
Detrend Price = Actual Price - (45,600 +23.2*Year)
What is commong scaling - ANSWER-data between 0 and 1
which is linearly
what is standard scaling - ANSWER-scaling to normal
distribution: mean of 0, std = 1; standardization
when to use type of scaling method? - ANSWER-Data in a
bounded range is good for scaling like: neural networks, SAT
score;
Data sometimes in standardization like PCA and clustering;
, 4|Page
K-Nearest Neighbor Algorithm - basic machine learning -
ANSWER-Find the class of a point: pick the k closest points to
the new one
The new point's class is the most common among the k
neighbors
For K-Nearest Neighbor keep in mind: - ANSWER-1. Can use
other distance metrics;
2. attributes can be weighted by importance
3. unimportant attributes can be removed.
4. Choose a good value of k
How do we evaluate the model output for SVM? - ANSWER-1.
look at coefficients of the predictors, if close to zero they are
probably not good.
2. C Value - up or down to trade off error rate and margin
3. Accuracy - data point accurately
CUSUM - ANSWER-short for cumulative sum; change
detection method that compares observed distribution mean with
a threshold level of change. C and T control values.
ISYE 6501 Midterm 1 Exam Newest 2025 Complete
100 questions and Correct Detailed Answers (Verified
Answers) |Already Graded A+
A survey of 25 people recorded each person's family size and
type of car. Which of these is a data point?
A. The 14th person's family size and car type
B. The 14th person's family size
C. The car type of each person - ANSWER-A.
observation/record of measurements for a single member of a
population or data set. In the standard tabular format, a row of
data.
The farther the wrongly classified point is from the line...... -
ANSWER-The bigger the mistake we've made
The term including the margin gets larger so the importance of a
large margin out weights avoiding mistakes and classifying
known data samples - ANSWER-As lambda gets larger
Each column is information about what? - ANSWER-An
observation
,2|Page
What is a Support Vector Macine? - ANSWER-Basic Machine
Learning; classifying a binary response based on a set of
attributes and data points.
In SVM, do we scale the data? - ANSWER-Must scale the data!
In SVM, what is C? - ANSWER-C Value or Cost - this is our
tradeoff factor between correctness and margin
What is Box Cox Transformation? - ANSWER-useful for
transforming a response to eliminate heteroscedasticity;
logarithmic transformation; stretches out the smaller range to
enlarge its variability; shrinks the larger range to reduce its
variability. The idea is to find the best value of lambda; first
check the Q-Q plot to see if you need to do it.
What is hteroscedasticity? - ANSWER-Unequal variance; might
cause bias in our data b/c variance at the upper end can make
estimation errors larger and push the model to fit those data
points better
,3|Page
When should detrending be considered? - ANSWER-Response,
Predictors, Factor-based model, regression, SVM, etc. to analyze
time series data
How to detrend - ANSWER-Factor by Factor; i.e.: simple linear
regression
Price = 45,600 +23.2 * Year
Detrend Price = Actual Price - (45,600 +23.2*Year)
What is commong scaling - ANSWER-data between 0 and 1
which is linearly
what is standard scaling - ANSWER-scaling to normal
distribution: mean of 0, std = 1; standardization
when to use type of scaling method? - ANSWER-Data in a
bounded range is good for scaling like: neural networks, SAT
score;
Data sometimes in standardization like PCA and clustering;
, 4|Page
K-Nearest Neighbor Algorithm - basic machine learning -
ANSWER-Find the class of a point: pick the k closest points to
the new one
The new point's class is the most common among the k
neighbors
For K-Nearest Neighbor keep in mind: - ANSWER-1. Can use
other distance metrics;
2. attributes can be weighted by importance
3. unimportant attributes can be removed.
4. Choose a good value of k
How do we evaluate the model output for SVM? - ANSWER-1.
look at coefficients of the predictors, if close to zero they are
probably not good.
2. C Value - up or down to trade off error rate and margin
3. Accuracy - data point accurately
CUSUM - ANSWER-short for cumulative sum; change
detection method that compares observed distribution mean with
a threshold level of change. C and T control values.
