Cross-Validation
Evaluation & MetricsChapter 10: Report Card for BitTerm 165 of 253
Testing the model on several different data splits for a reliable estimate.
At Sunrise School
Gopal asksIs one test split enough?
Bit answersCross-validation tests on 5 different splits and averages them. 89, 91, 90, 88, 92: reliable, not lucky.
In the real worldTaking multiple practice exams.
Watch the short
Short coming soonSeason 5 · Day 34Follow @haigopala
Short coming soonSeason 5 · Day 34Follow @haigopala.telugu
Short coming soonSeason 5 · Day 34Follow @haigopala.hindi
Quick check
Quick check. Is one lucky test score enough to trust a model? Yes or no?
No! Average several splits.
Taught together with
Practise it
Try this term and all 253 others in the free Model Training Arena.
Open the Arena →Follow Bit daily: YouTube English · తెలుగు · हिंदी | Instagram English · తెలుగు · हिंदी