Slides 2 - Software Metrics-Foundation

Question 1

Why measure?

Answer 1

• Quantifying experience (Erfahrung quantifizieren)
• Assessment (Bewertung)
• Prognosis (Prognose)
• > measure immaterial things

Question 2

What happens if we don’t measure?

Answer 2

• we cannot define measurable goals
  (keine messbare Ziele definieren)
  product maintainable and reliable?
  (Produkt wartbar und zuverlässig)
• we cannot analyze important cost drivers in detail
  (kostentreiber in Detail können nicht analysiert werden)
• we cannot quantify important qualitative properties of product or processes
  (Wir können wichtige qualitative Eigenschaften des Produkts nicht quantifizieren)
  How hight is the probabilty of failure?
  (Ausfallwahrscheinlichkeit)
• we cannot assess the efficiency of new technologies
  (Wir können die Effizienz neuer Technologien nicht beurteilen)

Question 3

Definition: Measurement

Answer 3

act or process of assigning a number or category to an entity to describe an attribute of that entity

(Handlung oder Prozess der Zuweisung einer Nummer oder Kategorie zu einer Entität, um ein Attribut dieser Entität zu beschreiben)

Question 4

Examples of Measurable Attributes

Answer 4

• Air temperature,
• Height of a person,
• Size of a t-shirt,
• Size of a program

Question 5

Definition: Model

Answer 5

• A model is an image or a representation of something.
• The modeled something is called original.
• It can be real or virtual.

Question 6

General Model Theory - property

Answer 6

• Mapping property (Zuordnungseigenschaft)
  ● It is a mapping of the original.
  (Abbildung des Orginals)
• Reduction property (Reduktionseigenschaft)
  ● It does not contain all attributes of the original.
  (enthält nicht alle Attribute des Orginals)
• Pragmatic property (Pragmatisches Eigentum)
  ● It serves for a specific purpose
  (Es dient einem bestimmten Zweck)

Question 7

Representational Theory of Measurement

Answer 7

● Measuring is modeling
● An attribute of the original is mapped to a numerical system.
(Ein Attribut des Orginals wird einem Zahlensystem zugeordnet)
● At first, it is mapped to an empirical system.

● Empirical relational system
Defines attributes of entities and their relations in the real world.
● Numerical relational system
Formalization of attributes and relations by means of numbers or symbols

Question 8

Representational Condition of Measurement

Answer 8

1. Properties that hold in the empirical system
   must also hold in the numerical system.
2. If properties do not hold in the numerical
   system but in the empirical system, the applied
   metric is inappropriate

Question 9

Empirical Relation System

Answer 9

1. Select an attribute of entities.
2. Identify and define relations that hold for this attribute

ERS
E: set of entities
ER: set of empirical relations

Question 10

Numerical Relation System

Answer 10

A numerical relation system to map the values of the attributes.
(Ein numerisches Beziehungssystem, um die Werte der Attribute abzubilden)

NRS
N: set of numbers
NR: set of numerical relations

Question 11

Representational condition

Answer 11

If a relation r ϵ ER holds for attribute A, then the

corresponding relation r’ ϵ NR must hold as well

Question 12

Key Steps of Formal Measurements

Answer 12

• Identify relevant attribute for real-world entities
• Identify empirical relations for this attribute
• Identify corresponding numerical relations
• Define a mapping from real-world entities to numbers
• Check that the representational condition holds

Question 13

Kinds of Scale

Answer 13

Nominal Scale
● Pure classification of values
● Classes are not ordered

 Ordinal Scale
● Values are totally ordered. 
● Values represent a ranking
● Arithmetic operations have no meaning
● Median value can be computed.
(Mittelwert kann berechnet werden)

Interval Scale
● Captures information about the size of the intervals that separate the classification
● Distance can be computed and are comparable
● Mean value exists
(Mittelwert vorhanden)

Rational Scale
● Values are totally ordered and additive
● Percentage values are computable

Absolute Scale
● Values are absolute entities
● Measurement is made by counting

Question 14

What can be measured?

Answer 14

Product qualities
● Product qualities
● robustness
● usability
● testability
● efficiency

Process qualities
● duration
● effort
● planning precision

Question 15

Classification - Source of a Metric

Answer 15

Model-Based metric:
Combines directly measurable or estimated
attributes according to a given formula

Empirical metric:
Based on observations in the real world, without an
underlying model.
(Basierend auf Beobachtungen in der realen Welt, ohne eine zugrundeliegendes Modell)

Question 16

Classification - Manipulation

Answer 16

Robust metric:
● Returns results that can not be manipulated if the metric is known

Undermineable metric:
● Returns authentic results only if the application of the metric is not known

Question 17

Quality Properties of Metrics

Answer 17

Plausibility:
● Does the metric really measure the selected attribute?
● Representation condition: does the value represent the attribute?

Relevance:
● Are the values useful?

Profitability:
● How much does the measurement cost?
● Are the costs of a measurement justified by the relevance?

Comparability & Precision: (Vergleichbarkeit & Genauigkeit)
● Which analyses can be done based on the values?
● For comparison at least the ordinal scale has to be used
● Are different attribute states mapped on different values?

Availability:
● Can an attribute be measured when the values are needed?

Reproducibility:
● Do multiple measurements of the same attribute (through different people / in different places) produce the same values?