システム開発プロジェクトにおいて, 成果物の品質を評価するために使用する指標として, 適切なものはどれか。
平成22年度 秋期 ITパスポート試験 問51
This question's explanation and your answers so far.
問 51
Incorrectア
The external procurement rate expresses the proportion of the resources needed for development (personnel, parts, software components, and so on) that are procured from outside rather than supplied in-house. It is a metric showing the procurement policy and the development structure and cost makeup; whether the ratio is high or low, that alone does not tell you whether the deliverable turned out well or badly. It is not a metric for evaluating quality.
- イ
正解 The correct answer is (イ) Test coverage rate. Test coverage rate is the proportion of the statements or branches (conditions) in a program that are actually executed and verified by testing. In other words, it expresses numerically how much of the deliverable that was built has been verified (the degree of coverage), so it directly evaluates how soundly the deliverable itself was made — that is, its quality. If coverage is low, many unverified parts have never been run even once, so it can be judged that defects are likely hidden there. Conversely, if coverage is high, it can be judged that defects have been sufficiently found by testing. Because it thus provides a basis for judging whether quality is good or bad, it is appropriate as a metric used to evaluate the quality of deliverables.
Incorrectウ
The total number of people deployed is a metric expressing the size of the effort in units such as person-months or person-days, and it is used for cost management and for analyzing productivity (amount of deliverables divided by effort). Putting in many people does not necessarily mean quality is high, and the reverse holds as well, so it cannot serve as a yardstick for measuring whether quality is good or bad.
Incorrectエ
The number of days elapsed in the project is a time- and schedule-related metric expressing how much time has passed since the start, and it is used to check progress management and adherence to the delivery deadline. It shows how far the work has advanced, but says nothing about the quality of the deliverable that has been produced.
- Sort the metrics by the QCD viewpoints (Quality, Cost, Delivery) before choosing. Quality (Q) = test coverage rate, defect density, number of issues raised in reviews; Cost (C) = effort put in, total number of people; Delivery (D) = days elapsed, progress rate. The standard approach is to identify which viewpoint the question asks about and then pick the metric that matches it.
- Definition of test coverage rate (coverage): the proportion of the statements or branches executed during testing out of the total. The higher the coverage, the fewer the parts left unverified and the more reliable the detection of defects by testing, which is why it is used as a quality metric for deliverables.