Data Warehousing Questions and Answers

Answer: B

Correct answer: Current data intended to be the single source for all decision support systems.

Reconciled data is the detailed, current, integrated layer that results from extracting and cleaning the operational sources - the enterprise data warehouse itself. It is normalised and not usually queried directly; derived data, formatted for particular user applications, is what the data marts hold.

Answer: C

Correct answer: Completely normalized

The fact table holds the measures and the foreign keys to the dimensions, and nothing that could be moved elsewhere - so it is fully normalised. It is the dimension tables that are deliberately denormalised, since flattening them reduces the joins a query must perform, at a cost in redundancy that a read-only store can afford.

Answer: A

Correct answer: True

Status data record what something is at a point in time - a balance, an address, a price. Event data record that something happened - a payment, a change of address. Both are held, and the distinction matters because a status can often be reconstructed from a sequence of events, and vice versa.

Answer: A

Correct answer: False

Detailed, current data intended as the single authoritative source is reconciled data - the warehouse itself. Derived data is what has been selected, aggregated and formatted for particular end-user applications, and it is what the data marts hold.

Answer: A

Correct answer: True

Scrubbing improves the data going into the warehouse but leaves the source systems producing the same errors next time, so the cleaning must be repeated at every load. The lasting remedy is to correct the operational processes, which is why quality is treated as a management programme rather than a stage in ETL.

Answer: D

Correct answer: All of the above

A multifield transformation is one whose input or output spans more than one field, and all three patterns qualify - splitting a full name into parts, combining separate address lines into one, or mapping several fields onto several others. What distinguishes it is the crossing of field boundaries, not the direction.

Answer: A

Correct answer: To explain some observed event or condition

Data mining is exploratory: it looks for patterns and relationships that were not anticipated, and so can suggest why something happened. This is what distinguishes it from ordinary querying, which confirms relationships already expected - the second and third options describe verification rather than discovery.

Answer: B

Correct answer: False

Joining combines data from several sources into a single record; the process described - partitioning according to predefined criteria - is selection. Both are transformation operations, and the statement has attached the definition of one to the name of the other.

Answer: C

Correct answer: Organized around important subject areas.

Subject orientation is the first of the defining characteristics: the warehouse is arranged around customer, product or sale rather than around the applications the data came from. The other options describe an operational system - it is the warehouse that is integrated, non-volatile and time-variant.

Answer: A

Correct answer: True

A surrogate key is a meaningless integer generated for the warehouse, and using one insulates the schema from the operational keys, which may be reused, may change format, or may collide between source systems. It also makes possible the slowly changing dimension, where several rows describe the same real entity at different times.