Answer-first summary for fast verification
Answer: No
The solution does not meet the goal because stratified sampling only ensures that the class proportions in the training and test sets mirror the original dataset's imbalance; it does not address the imbalance itself. The goal is to 'compensate for the class imbalance,' which requires techniques that modify the dataset to balance class distribution, such as oversampling (e.g., SMOTE), undersampling, or using class weights. Community discussion strongly supports this, with high-upvoted comments noting that stratified split 'does not guarantee balanced dataset' and that SMOTE or similar methods are needed to generate or adjust data for imbalance compensation. While stratified sampling is useful for maintaining representativeness in splits, it does not resolve the underlying imbalance issue.
Author: LeetQuiz Editorial Team
Ultimate access to all questions.
You are creating a new experiment in Azure Machine Learning Studio. The training set has a significant class imbalance, with one class having far fewer observations than the others. You need to choose a suitable data sampling strategy to address this imbalance.
Solution: You use the Stratified split for the sampling mode.
Does this solution meet the goal?
A
Yes
B
No
No comments yet.