Database Processing for BIS Questions and Answers

Answer: C

Correct answer: Do what-if analysis only

Data mining applies statistical and mathematical techniques to make predictions and support what-if analysis. RFM belongs to reporting, since it uses only sorting, grouping and simple arithmetic, and transaction processing belongs to operational systems. The boundary between reporting and data mining is drawn exactly at the sophistication of the technique.

Answer: A

Correct answer: frequently

The three letters stand for recency, frequency and monetary value - how recently the customer last ordered, how often they order, and how much they spend. Each is scored one to five by dividing the customer base into fifths, and the resulting three-digit score is what the analysis ranks customers by.

Answer: B

Correct answer: analyze and rank customers

RFM analysis scores each customer on how recently they ordered, how frequently they order and how much money they spend, and uses the three scores to rank them into groups. It is a reporting technique built on simple sorting and grouping, not a predictive one, which is why it belongs with reporting rather than with data mining.

Answer: A

Correct answer: True

The customer base is divided into fifths on each of the three measures, and a customer receives a score of 1 to 5 on each according to which fifth they fall into, 1 being the best. The three scores together form a code such as 1 1 5 or 5 5 5, and it is that code by which customers are grouped and acted on.

Answer: A

Correct answer: Cluster analysis only

Unsupervised techniques are applied without a hypothesis or a target variable: the analyst looks for patterns and then constructs an explanation for whatever emerges. Cluster analysis is the type example. Regression is supervised, since a model is specified in advance and fitted, and RFM is a reporting technique rather than data mining at all.

Answer: B

Correct answer: False

The description given - sums, averages and groupings - is reporting, not data mining. Data mining applies statistical and mathematical techniques such as regression, clustering, decision trees and neural networks. The line between the two branches of business intelligence is drawn exactly at the sophistication of the technique used.

Answer: A

Correct answer: True

The four report types are static, produced once and unchanged thereafter; dynamic, generated at the time of the request from current data; query, where the user supplies criteria; and OLAP, where the user manipulates measures against dimensions and drills through levels. The four differ in how much control the reader has over what is produced.

Answer: B

Correct answer: False

Warehouse data is usually denormalised, not normalised. Normalisation protects a database against modification anomalies, which matters for operational data that is constantly updated; a warehouse is loaded in batches and read heavily, so the joins that normalisation requires cost more than the redundancy it would avoid. Star schemas are the usual result.

Answer: A

Correct answer: True

Authoring a report covers everything from connecting to the data sources and writing the query to laying out the result and formatting it for presentation. The other two functions of a report management system are management, which controls who receives what and when, and delivery, which sends it out.

Answer: C

Correct answer: RFM analysis only

RFM classifies customers into groups from their purchasing behaviour using nothing more than sorting and simple arithmetic, which places it firmly in reporting. Cluster analysis and regression are data mining techniques, requiring statistical machinery that a reporting system does not provide - the distinction the whole chapter turns on.