Free CompTIA DataX DY0-001 Actual Exam Questions - Question 8 Discussion
features more useful when building a model. Which of the following data transformation techniques
should the data scientist use? (Choose two.)
B/D? One-hot avoids any rank assumption, while label encoding turns categories into numbers for models that can handle ordinal data. Other choices like normalization don’t fit categorical variables at all.
B and D. One-hot encoding is great for nominal categories to avoid implying order, while label encoding can work if the categories have a meaningful order. The other options either don’t apply to categorical data or are more about scaling numerical features.
It’s B and D. Normalization and scaling are for continuous variables, so A and E don’t fit here. Pivoting is more about restructuring data tables, not encoding categories. Linearization isn’t really a standard term in this context. One-hot encoding (B) turns categories into binary vectors, which works well for nominal data, while label encoding (D) assigns integers to categories, useful for ordinal data or when the model can handle categorical integers. So these two cover the usual approaches to preparing categorical features for modeling.
B and D, as those are the standard ways to convert categorical data for models.
B and D, since one-hot and label encoding both directly transform categorical features.
B imo for turning categories into binary flags, and D because assigning numbers is quick, though not perfect. Normalization and scaling don’t make sense here since they’re for continuous data.
It’s B and D. Neither normalization (A) nor scaling (E) really help with categories since those are for numeric data. Linearization (C) isn’t a standard term for categorical encoding either. Pivoting (F) changes table format but doesn’t turn categories into numbers for models. So one-hot (B) and label encoding (D) are the two solid picks here.
It’s definitely B and D. One-hot encoding creates binary columns and avoids any implied order, while label encoding just assigns integers but can mislead some models if treated as ordinal data. Pivoting is more about data structure and not really about converting categorical features into model-friendly formats. Normalization and scaling don’t make sense here since they’re for numerical data, so they’re off the table. Linearization isn’t a common term for handling categorical variables, so that one’s unlikely too.
B/D for sure. Also, label encoding is simple but can add unintended order, so one-hot encoding helps avoid that. Scaling and normalization don’t apply to categorical features, so they’re out.
B and D again, since pivoting changes shape, not feature type.
Maybe B and D are right because one-hot encoding keeps categories separate, and label encoding assigns numbers. Pivoting is more about reshaping tables, not really transforming features for models.
I agree B and D fit best since normalization and scaling are for numerical data. But what about pivoting (F)? Could that help reshape the data for certain algorithms, or is it just not relevant here?
B and D make sense here—one-hot and label encoding are classic for categorical data. Normalization and scaling don’t really apply to categories.