DM week 10
- Created by: mariolarc
- Created on: 14-05-21 10:59
Other questions in this quiz
2. Consider a classification problem with one target attribute and categorical input attributes. Which of the following is the least relevant for effectively performing feature selection to solve this classification problem?
- Remove an input attribute if it is highly correlated with another input attribute.
- Remove an input attribute x if the mutual information between the target attribute and x is low.
- Remove input attributes with a small set of possible values
- Remove input attributes which are independent of the target attribute.
3. Which of the following statements about principal component analysis is false?
- Identifies directions with low variance.
- Two different principal components are orthogonal to each other.
- The number of principal components is as most equal to the number of features.
- Principal component analysis is a linear transformation.
4. Which of the following ways of transforming numeric variables is particularly useful for improving the performance of data mining methods which are based on distance metrics?
- Conversion into percentiles.
- Converting them into binary indicator variables.
- Standardization.
- Binning (to replace numeric variables with categorical ones).
5. Backward feature selection is typically:
- A practical approach.
- A time-consuming approach.
- A probabilistic approach.
- A regression approach.
Comments
No comments have yet been made