Input data risks · Question 23 of 60
Against which risks does data representativeness testing help?
3 answers are correct. Medium
Show answer
Answer
- A Dataset skewed through data augmentation
- C Some target classes appear far too rarely in the dataset
- D The training data reflects only some of the use cases
Explanation
Representativeness testing helps against imbalanced datasets, datasets skewed by augmentation, missing data, training data focused on a subset of use cases and value ranges not covered. Wrong data types are a case for dataset constraint testing, and ambiguous data one for label correctness testing.