Core data concepts
Start with the question the data must answer, not a file extension. Compare regular CSV columns with JSON documents whose keys may vary, and with images or audio that require extraction before their content can be counted. Separate interchange formats from object storage and an operational database: retaining proof images differs from updating an order state or preventing a double sale. Concurrent state changes indicate a transactional workload; grouping history by month or category indicates analytical work. In the extended case, trace record counts from source to pipeline, model and visual. The first count divergence identifies where to investigate; blaming the chart alone is insufficient. Distinguish the roles that verify source data, repair transformations and agree on metric meaning. Exercises classify records, calculate a rate and reconcile a pipeline, while explicitly marking what cannot be inferred from unindexed scans or incomplete fields.
Read the whole first subject without an account: its lessons, summary and review questions.
Subject lessons
- Lesson 1 Representing data: structure is not a file extension
- Lesson 2 From files to a store that fits the query
- Lesson 3 Distinguishing transactions from analysis
- Lesson 4 Who owns which decision in the data chain?
- Lesson 5 Read file formats before choosing a store
- Lesson 6 Data quality and lineage across roles
- Integrated subject summary Review questions
The remaining subjects unlock in sequence once you create a free account.