MSCI 446 Midterm Practice Questions with Correct Answers Latest Update 2025-2026
What is unsupervised machine learning? - Answers -Explain the past - at face value, what does it
tell you?
o Not interested in learning the future
-Segmentation, association
What is supervised machine learning? - Answers -Predict the future based on the past given
data
-Predict the value of some variable (past) using the values of other variables
How does email spam filtering work? - Answers Input: examples of legitimate emails and spam
emails
Output: a program that predicts whether a given email is legitimate or spam - could be very bad
depending on how general the learning is (i.e. keywords vs. exact sentences)
Explain One Hot Encoding - Answers Categorical -> Boolean/Binary: If the categorical variable
has k values, one hot encoding creates k Boolean variables, one of which will be true in any one
record
Explain Document Term Vectors - Answers o A fragment of vector from a short document with
text
o Unlocks machine learning algorithms that previously required numeric values
o Consequence: Fails to recognize words that have similar semantics because they have
different spellings
o Bert model: word embeddings
What is an outcome? - Answers · The variable we want to predict
· If the outcome variable is categorical, we say we are doing classification
· Otherwise, we say we are doing numeric prediction
· Classifier: The algorithm for classification
What is a predictor? - Answers · The variables we will use to predict the outcome variable
· Also referred to as explanatory variable or features
What is a good feature? - Answers A good feature should be correlated with the label it predicts
What is garbage-in-garbage-out? - Answers Data Quality
, o Duplicate data
o Incorrect data
o Missing data
What are some examples of garbage-in-garbage-out? - Answers a very high salary in a boxplot
of salaries -> owner/CEO is outlier,
employees age between 17 and 65,
definition of a 'trip' to track bikes,
GPS accuracy to within 100 meters,
salary on an hourly/weekly/annual basis,
temperature = 0,
default postal code
What do you do if a value in the data is missing? - Answers Take value from entry with most
similar variables (best choice) - distance
What is support of an itemset A? - Answers sup(A) or |A| is the number of transactions
containing all the items in A
What is an association rule? - Answers A statement of the form A -> B, where A and B are
itemsets
What is the support of A -> B? - Answers |AB|
What is the confidence of A -> B? - Answers |AUB|/|A|
For n sets, what is the range of support? - Answers 0 to n
For n sets, what is the range of confidence? - Answers 0 to 1
What is the computational complexity of the association rule mining problem? - Answers
Exponential (2^n)
What is a good minimum confidence? - Answers 0.8 - increment by 0.05 until this is reached
What are the steps in the Apriori algorithm? - Answers 1. Find all "frequent itemsets" with
support >= s
2. Using the frequent itemsets that meet the threshold, find all association rules with confidence
>= c
What is unsupervised machine learning? - Answers -Explain the past - at face value, what does it
tell you?
o Not interested in learning the future
-Segmentation, association
What is supervised machine learning? - Answers -Predict the future based on the past given
data
-Predict the value of some variable (past) using the values of other variables
How does email spam filtering work? - Answers Input: examples of legitimate emails and spam
emails
Output: a program that predicts whether a given email is legitimate or spam - could be very bad
depending on how general the learning is (i.e. keywords vs. exact sentences)
Explain One Hot Encoding - Answers Categorical -> Boolean/Binary: If the categorical variable
has k values, one hot encoding creates k Boolean variables, one of which will be true in any one
record
Explain Document Term Vectors - Answers o A fragment of vector from a short document with
text
o Unlocks machine learning algorithms that previously required numeric values
o Consequence: Fails to recognize words that have similar semantics because they have
different spellings
o Bert model: word embeddings
What is an outcome? - Answers · The variable we want to predict
· If the outcome variable is categorical, we say we are doing classification
· Otherwise, we say we are doing numeric prediction
· Classifier: The algorithm for classification
What is a predictor? - Answers · The variables we will use to predict the outcome variable
· Also referred to as explanatory variable or features
What is a good feature? - Answers A good feature should be correlated with the label it predicts
What is garbage-in-garbage-out? - Answers Data Quality
, o Duplicate data
o Incorrect data
o Missing data
What are some examples of garbage-in-garbage-out? - Answers a very high salary in a boxplot
of salaries -> owner/CEO is outlier,
employees age between 17 and 65,
definition of a 'trip' to track bikes,
GPS accuracy to within 100 meters,
salary on an hourly/weekly/annual basis,
temperature = 0,
default postal code
What do you do if a value in the data is missing? - Answers Take value from entry with most
similar variables (best choice) - distance
What is support of an itemset A? - Answers sup(A) or |A| is the number of transactions
containing all the items in A
What is an association rule? - Answers A statement of the form A -> B, where A and B are
itemsets
What is the support of A -> B? - Answers |AB|
What is the confidence of A -> B? - Answers |AUB|/|A|
For n sets, what is the range of support? - Answers 0 to n
For n sets, what is the range of confidence? - Answers 0 to 1
What is the computational complexity of the association rule mining problem? - Answers
Exponential (2^n)
What is a good minimum confidence? - Answers 0.8 - increment by 0.05 until this is reached
What are the steps in the Apriori algorithm? - Answers 1. Find all "frequent itemsets" with
support >= s
2. Using the frequent itemsets that meet the threshold, find all association rules with confidence
>= c
