Learn Before
Explain why informal variance requires comparing two kinds of performance.
Question: In a concise analytical response, explain why training performance alone or dev/test performance alone is insufficient to express the supplied informal definition of variance.
Sample answer: Informal variance is defined as how much worse an algorithm does on the dev or test set than on the training set. Therefore, it depends on a comparison between training performance and dev or test performance. Either result alone omits the gap that the term variance informally describes.
Key points:
- Training performance is the reference.
- Dev or test performance is compared with training performance.
- Variance is the amount by which dev or test performance is worse.
- A single performance result does not express the comparison.
Rubric: A strong response accurately identifies both sides of the comparison, explains that variance is the amount of worsening between them, and states why either performance result alone is insufficient.
0
1
Tags
Machine Learning
Deep Learning
Supervised Learning
Dive into Deep Learning @ D2L
Data Science
Machine Learning Strategy
Machine Learning Yearning @ DeepLearning.AI
Related
Which comparison informally describes an algorithm's variance?
Variance informally concerns a dev-or-test versus training performance gap.
Variance is how much worse performance is on the test set than on the _____ set.
Match each term to its role in the informal definition of variance.
Order the reasoning steps used to identify informal variance.
Explain why informal variance requires comparing two kinds of performance.
Diagnose what a training-to-dev performance decline represents.
What does the word “worse” indicate in the informal variance comparison?
Which observation supplies the information needed to interpret variance informally?
Reversing the training and test sets preserves the informal variance definition.