Database Processing for BIS Questions and Answers

Answer: B

Correct answer: False

Market basket analysis rests on more than conditional probability. Support is the unconditional probability that a set of items occurs together, confidence is the conditional probability of one given another, and lift compares the two to show whether the association is real. Describing the technique as primarily conditional overstates one of the three.

Answer: C

Correct answer: use statistical procedures to predict future events

A reporting system creates information from existing data by filtering, sorting, grouping and calculating simple arithmetic, and then delivers it to the right users at the right time, drawing on whatever sources are needed. Prediction using statistical procedures belongs to the other branch of business intelligence, data mining, which is exactly the distinction the two categories are defined by.

Answer: B

Correct answer: False

The difference between a data mart and a data warehouse is precisely one of scope. A warehouse serves the whole organisation; a data mart is smaller and addresses a single business function or department, drawing its data from the warehouse. Saying it still addresses the entire business removes the distinction.

Answer: C

Correct answer: Data metadata

The components of a data warehouse are the extraction, cleaning and preparation programs, the warehouse DBMS, the warehouse data itself, and the warehouse metadata that describes it. “Data metadata” is not a term in that list - the metadata is the warehouse’s metadata, describing where the data came from, what it means and how current it is.

Answer: C

Correct answer: Too much data

The curse of dimensionality is the problem of having so many attributes that the analysis becomes unmanageable: the number of possible combinations grows exponentially with the number of dimensions, and the data available thins out across them. It is a problem of quantity and breadth rather than of quality, so it is distinct from dirty, inconsistent or non-integrated data.

Answer: A

Correct answer: True

This is what makes the technique unsupervised: no model or hypothesis is specified in advance, so the analyst applies the technique, sees what patterns emerge, and only then constructs an explanation for them. Supervised mining works the other way, specifying a model first and then fitting it to estimate its parameters.

Answer: B

Correct answer: Pull

The distinction is who initiates the delivery. In pull mode the user goes to the website and requests the report; in push mode the system sends it out on a schedule or when an event occurs, by e-mail or to a dashboard. A report sitting on a site waiting to be accessed is therefore pull, whatever technology delivers it.

Answer: A

Correct answer: True

The name is a misnomer because a neural network in data mining is not a model of the brain and has no biological content: it is a nonlinear regression technique, a set of weighted functions fitted to data. The name reflects the historical origin of the idea rather than what the technique actually does.

Answer: A

Correct answer: True

OLAP is a reporting technique: it summarises existing data by arranging measures against dimensions and lets the user drill through levels, but it makes no prediction and applies no statistical model. It is more interactive than a static report, which is why it is treated as its own report type, but it remains reporting rather than data mining.

Answer: B

Correct answer: support

Support is the probability that a set of items appears together in a transaction - here the fraction of the thousand rentals containing both Video C and Video D. It is an unconditional probability of the whole combination. Confidence, by contrast, is conditional: the probability of D given that C was rented.