CPA Advanced Level

Business Data Analytics flashcards

Read the question, say the answer to yourself, then turn the card to check. The first 10 of 151 cards are free; the rest come with the subject.

Card 1 of 151

Jump to a card

Complete Flashcard Reference

Review every card in this deck. Open any card to reveal its answer. The first 10 are free; the other 141 come with the subject.

What are the five Vs of big data, and what does each mean?

Volume: the sheer size of the data (terabytes, petabytes). Velocity: the speed at which it is generated and processed, up to real-time streams. Variety: the many types, formats and sources. Veracity: its accuracy, trustworthiness and quality. Value: the usefulness of the insight for decisions, the only reason to collect it.

What are the six phases of CRISP-DM, in order?

1. Business understanding 2. Data understanding 3. Data preparation 4. Modelling 5. Evaluation 6. Deployment The model is iterative: data understanding often sends the team back to the business question, modelling back to preparation, and evaluation can restart the cycle.

Which question does each of the four types of analytics answer?

Descriptive: what happened? (management accounts, ratios) Diagnostic: why did it happen? (variance analysis, drill-down) Predictive: what is likely to happen? (forecasts, credit scoring) Prescriptive: what should we do? (Solver, optimisation) Value and difficulty rise in that order.

What do A1, $A$1, $A1 and A$1 do when a formula is copied?

A1: column and row both move. $A$1: neither moves; use it for a single input such as a discount rate. $A1: the column is fixed, the row moves. A$1: the row is fixed, the column moves. F4 cycles a reference through the four forms while you edit it.

How does the argument order of SUMIF differ from SUMIFS?

SUMIF(range, criteria, [sum_range]): the test range first, the range to add last and optional. SUMIFS(sum_range, criteria_range1, criteria1, ...): the range to add first. COUNTIF(range, criteria) also puts the range first. Writing SUMIFS even for one condition makes a second condition easy to add.

How do conceptual, logical and physical data models differ?

Conceptual: the main entities and their relationships, in business terms, with no technology. Logical: every entity's attributes, primary and foreign keys and cardinality, normalised, still independent of any database product. Physical: tables, exact data types, indexes and constraints for one DBMS. They are built in that order, then implemented.

What are the three defensible ways to handle missing values?

Delete: drop the records, safe when few are affected and they are missing at random. Impute: estimate from other observations (the mean, or the median, which resists outliers) and mark the imputed values. Flag: keep the record, tag the gap and exclude it only from calculations it would distort. Document the choice in the cleaning log.

What are the four syllabus categories of visualisation, with a typical chart for each?

Comparison: column or bar chart. Composition (part of a whole): pie, stacked column, waterfall. Distribution (spread of one variable): histogram, box and whisker. Relationship (two or more variables together): scatter, bubble.

How do supervised, unsupervised and reinforcement learning differ?

Supervised: learns from labelled examples where the answer is known (loans marked default or repaid); classification and regression. Unsupervised: finds structure in unlabelled data, such as customer clusters or anomalies. Reinforcement: learns by trial and reward, as in dynamic pricing.

Why is the year 0 outlay added outside Excel's NPV function?

NPV assumes its first value arrives one period from now, so it discounts every value it is given. Correct: =NPV(rate, years 1 to n) + year 0. Putting year 0 inside the range discounts the whole answer by one extra period: the result equals the true NPV / (1 + rate).

What does XLOOKUP do that VLOOKUP cannot?