BCOR 2205 Quizzes Questions
with All Correct Answers
Which of the following is/are characteristics of Auto ML tool like Data Robot? - Answer--
Models will be ranked
-Models can be run simultaneously
Which of the following is the most important criterion of Machine Learning? - Answer-
Accuracy
Unsupervised machine learning does not have a pre-set target: - Answer-True
When building an auto machine leaning model, you always use both internal and
external data. - Answer-False
Which of the following is/are part of the success criteria of our model? - Answer--who
uses the model
-how much value the model can drive
If you are a phone company and you are trying to determine whether a person will
upgrade their phone or not, the unit of analysis could be : - Answer-Person- with data
like Customer ID, Gender, Zip code, Income, Phone Brand, and the target is Phone
Upgrade
Which of the following is/are a criteria that should be used to evaluate any proposed
project? - Answer--Is the project statement in the language of the business?
-How could solving the problem improve the bottom line?
-Does the project specify actions that should result from the project?
Which of the following is NOT part of the Define Project Objectives stage in the Machine
Learning Life Cycle Model? - Answer-find the appropriate data
We don't need to consider ethical risks when developing our machine learning model. -
Answer-False
Which of these is not an example of a data source businesses typically use? - Answer-
SME?
Data used by your organization that is not collected by your organization is called
__________________? - Answer-External Data
, attribute describe entities - Answer-true
Using personal and business insight to change the features is Target Leakage -
Answer-False
Data Leakage and Target Leakage are synonymous: - Answer-true
The foreign key must be unique and non-null. - Answer-False
Acquire, Explore, and Manipulate Data is the first step in the Machine Learning life cycle
discussed in class and the textbook: - Answer-False
If you were to remove a feature where you would not have the value at the time you'd
implement the model to make the prediction, you would be doing so because of -
Answer-Target Leakage
The three types of Dirty data mentioned in class are inaccurate information, duplicated
information, and missing information. - Answer-True
Compressing 'median income of the community' into ranges would be an example of
Feature Engineering: - Answer-True
In One-Hot-Encoding we code items that satisfy the criterion as 1 and items that do not
as 0. - Answer-true
It is easy to beat Data Robots accuracy score using Feature Engineering - Answer-
False
Which of the following are example of Target Leakage (select all that apply): - Answer-
???
Assume that we have 100 rows of data and we want to perform a five-fold cross
validation with a 20% holdout.
What will be the size of the holdout and each fold? - Answer-The holdout will be 20
rows, each fold will be 16 rows
The goal of machine learning is to build models with ______________________. -
Answer--High prediction capabilities
-High generalization capabilities
In k-fold cross validation we partition all the data into k equal subsets (including the
holdout): - Answer-False
In 5-fold cross validation each fold has a turn being the validation set while the other 4
folds are the training set: - Answer-True
with All Correct Answers
Which of the following is/are characteristics of Auto ML tool like Data Robot? - Answer--
Models will be ranked
-Models can be run simultaneously
Which of the following is the most important criterion of Machine Learning? - Answer-
Accuracy
Unsupervised machine learning does not have a pre-set target: - Answer-True
When building an auto machine leaning model, you always use both internal and
external data. - Answer-False
Which of the following is/are part of the success criteria of our model? - Answer--who
uses the model
-how much value the model can drive
If you are a phone company and you are trying to determine whether a person will
upgrade their phone or not, the unit of analysis could be : - Answer-Person- with data
like Customer ID, Gender, Zip code, Income, Phone Brand, and the target is Phone
Upgrade
Which of the following is/are a criteria that should be used to evaluate any proposed
project? - Answer--Is the project statement in the language of the business?
-How could solving the problem improve the bottom line?
-Does the project specify actions that should result from the project?
Which of the following is NOT part of the Define Project Objectives stage in the Machine
Learning Life Cycle Model? - Answer-find the appropriate data
We don't need to consider ethical risks when developing our machine learning model. -
Answer-False
Which of these is not an example of a data source businesses typically use? - Answer-
SME?
Data used by your organization that is not collected by your organization is called
__________________? - Answer-External Data
, attribute describe entities - Answer-true
Using personal and business insight to change the features is Target Leakage -
Answer-False
Data Leakage and Target Leakage are synonymous: - Answer-true
The foreign key must be unique and non-null. - Answer-False
Acquire, Explore, and Manipulate Data is the first step in the Machine Learning life cycle
discussed in class and the textbook: - Answer-False
If you were to remove a feature where you would not have the value at the time you'd
implement the model to make the prediction, you would be doing so because of -
Answer-Target Leakage
The three types of Dirty data mentioned in class are inaccurate information, duplicated
information, and missing information. - Answer-True
Compressing 'median income of the community' into ranges would be an example of
Feature Engineering: - Answer-True
In One-Hot-Encoding we code items that satisfy the criterion as 1 and items that do not
as 0. - Answer-true
It is easy to beat Data Robots accuracy score using Feature Engineering - Answer-
False
Which of the following are example of Target Leakage (select all that apply): - Answer-
???
Assume that we have 100 rows of data and we want to perform a five-fold cross
validation with a 20% holdout.
What will be the size of the holdout and each fold? - Answer-The holdout will be 20
rows, each fold will be 16 rows
The goal of machine learning is to build models with ______________________. -
Answer--High prediction capabilities
-High generalization capabilities
In k-fold cross validation we partition all the data into k equal subsets (including the
holdout): - Answer-False
In 5-fold cross validation each fold has a turn being the validation set while the other 4
folds are the training set: - Answer-True
