Data Quality
Domain: Data & Datasets
🎯 Motivation
How good the data is: accurate, complete, diverse, relevant, unbiased. Better data = better models.
📋 Concrete Example
Quality check: 95% accurate labels, 0% duplicates, diverse topics, no bias, relevant to task. High quality. vs 60% accurate labels, 20% duplicates, biased. Low quality.
🔗 Analogy
Ingredients for cooking - fresh, high-quality ingredients make better dishes. High-quality data makes better models. Data quality is the freshness of the ingredients.