Data Types and Data Quality for AI
Explore the differences between structured and unstructured data sources in AI projects, and understand the five key dimensions of data quality. This lesson helps you evaluate data readiness by assessing accuracy, completeness, consistency, timeliness, and representativeness to ensure AI investments are based on credible, reliable evidence.
The complaint data contains two valuable but different sources of evidence. The ticket database records each complaint in fields such as ticket ID, received date, assigned category, region, resolution time, and refund amount. The email archive and call transcripts contain the customer’s narrative, tone, product details, and the reasons behind the complaint. Before selecting an AI approach, we need to identify what kind of data each source provides and whether it is reliable enough to support the intended outcome.
Structured and unstructured complaint data
Structured data is information organized into consistent fields, usually rows and columns. In the complaint pilot, a row might represent one ticket, while columns hold values such as ticket_id, date_received, category, region, and refund_amount. Because the same field has the same meaning across records, structured data supports ...