Data Analytics Fundamentals
The Data Analytics Fundamentals domain tests whether you can classify data, distinguish analytics types, and judge data quality before analysis starts. Expect definition and scenario questions, not tool-specific commands.
Data, information, knowledge
Data are raw recorded facts or symbols without context. Information is data processed into meaning, such as a monthly sales total. Knowledge is information combined with experience and rules so it can guide action; wisdom adds judgment about when to act.
Analytics types
- Descriptive analytics answers what happened, using summaries, dashboards and KPIs.
- Diagnostic analytics asks why it happened, using drill-down, correlation and root-cause analysis.
- Predictive analytics estimates what is likely to happen, using statistical models or machine learning.
- Prescriptive analytics recommends what to do, using optimization, simulation or decision rules.
Data structures and quality
Structured data fit tables; semi-structured data have tags or hierarchy such as JSON; unstructured data include text, images and audio. Measurement scales matter: nominal data are labels, ordinal data are ordered labels, interval data have equal distances but no true zero, and ratio data have a true zero. Common data-quality dimensions are accuracy, completeness, consistency, timeliness, validity and uniqueness. Big data is often described by volume, velocity, variety, veracity and value. Data can also be primary or secondary; secondary data may not fit the original question.
Worked example
A retailer’s sales fall 12%. A descriptive dashboard shows the drop by month. Diagnostic drill-down by region finds stockouts in two warehouses. A predictive model estimates next quarter’s demand, and a prescriptive rule sets reorder points. Reasoning: each question changes the analytics type; a dashboard alone does not explain cause.
Traps
Confusing predictive with prescriptive is common: a forecast is not a recommendation. Do not treat ordinal labels as interval numbers, and do not assume volume guarantees value or quality. More data does not remove sampling bias, and correlation does not prove causation.