Isaca AAIA Exam - Topic 3 Question 13 Discussion

Actual exam question for Isaca's AAIA exam

Question #: 13
Topic #: 3

[All AAIA Questions]

When converting data categories before training an AI model, which of the following scenarios represents the GREATEST risk?

AOne-hot encoding the data attribute car colors for the options red, blue, green, black, white

BCreating dummy variables for the data attribute dog breed for the options labrador, terrier, beagle

COne-hot encoding the data attribute customer rewards category for the options economy, business, first class

DCreating dummy variables for the data attribute product flavor for the options vanilla, chocolate, strawberry, banana

Show Suggested Answer

Suggested Answer: C

The AAIA Study Guide emphasizes that encoding categorical variables must preserve the semantic meaning and order of categories when relevant. The greatest risk occurs when ordinal data---such as customer rewards tiers---is treated as nominal through one-hot encoding, which removes the inherent order and may impair model learning.

''Improper encoding of ordinal variables as nominal can distort the model's understanding of relationships, leading to inaccurate predictions or biased outcomes.''

Customer reward categories (economy < business < first class) have a natural order. One-hot encoding ignores this order, potentially degrading model accuracy. Other options represent nominal data and are appropriately encoded.

by Lavonna at Jan 05, 2026, 05:16 AM

Limited Time Offer

25%

Off

Get Premium AAIA Questions as Interactive Web-Based Practice Test or PDF

Contribute your Thoughts:

Submit Cancel

Wade

3 days ago

Hmm, I'd say B is the most dangerous. Dog breeds have a lot of hidden biases.

upvoted 0 times

...

Gail

8 days ago

D is the way to go. Dummy variables for product flavors are the safest bet.

upvoted 0 times

...

Melissia

14 days ago

I agree, C is the riskiest. Encoding customer rewards could reveal sensitive information about individuals.

upvoted 0 times

...

Emogene

19 days ago

Option C is the greatest risk. One-hot encoding customer rewards categories could lead to data leakage and overfitting.

upvoted 0 times

...

Marsha

24 days ago

I'm a bit confused about the differences between one-hot encoding and dummy variables. I wonder if that affects which scenario is riskier.

upvoted 0 times

...

Donte

29 days ago

I practiced a similar question, and I feel like creating dummy variables for dog breeds could introduce bias if some breeds are underrepresented.

upvoted 0 times

...

Deja

1 month ago

I think it might be option C, since customer rewards categories could have a significant impact on model performance if not handled correctly.

upvoted 0 times

...

Ryan

1 month ago

I remember discussing how one-hot encoding can lead to a high-dimensional space, but I'm not sure which option has the greatest risk.

upvoted 0 times

...

Alethea

1 month ago

I'm leaning towards the product flavor one-hot encoding as the riskiest. With that many unique flavors, you could end up with a ton of sparse columns that the model might have trouble generalizing from.

upvoted 0 times

...

Filiberto

2 months ago

For this type of data prep question, I usually try to think about the cardinality of the categories. The customer rewards one seems like it could have the most unique values, so I'd go with that as the highest risk.

upvoted 0 times

...

Shalon

2 months ago

Hmm, I'm a bit confused on this one. I was thinking the dog breed dummy variables might be the riskiest since there could be a lot of different breeds. But I'm not totally confident in that.

upvoted 0 times

...

Renato

2 months ago

I'm not totally sure about this one, but I think the greatest risk would be one-hot encoding the customer rewards category. That seems like it could lead to a lot of sparse data and potential overfitting.

upvoted 0 times

...