Data Fundamentals and Data Quality
What you must know
Candidates need a working grasp of data types (structured, semi-structured, unstructured), the data lifecycle from capture through storage, transformation and disposal, and the distinction between a data warehouse, a data lake and a lakehouse. Understand metadata (data about data), master data, reference data and transactional data, plus the concepts of primary keys, foreign keys and referential integrity. Know the common data quality dimensions: completeness, accuracy, validity, consistency, timeliness, uniqueness and integrity.
Traps in questions
Questions often blur validity (conforms to a defined format or rule) with accuracy (reflects the real-world fact). A date field can be perfectly valid yet wrong. Another trap is confusing a data lake (schema-on-read, raw storage) with a warehouse (schema-on-write, curated). Watch for questions implying more data automatically improves quality - volume does not fix bias or error. Sampling risk and completeness of the population are frequently tested in audit-flavoured items.
Rule of thumb
Before analysing, run a quality profile: count records, count nulls and duplicates, test ranges and formats, and reconcile totals to an independent source. Treat any unexplained variance as a finding until proven otherwise.
Worked example
A payroll file has 10,000 rows; 250 have a blank tax code and 40 duplicate employee IDs. Completeness is 97.5% (9,750/10,000) and uniqueness is 99.6% (9,960/10,000). Both are quantified, not assumed - report the metric, not a vague opinion.