Plan Metrics CS 406 Software Engineering I Fall 2001

Design metrics cs 406 software engineering i fall 2001 l.jpg
1 / 16
0
0
922 days ago, 350 views
PowerPoint PPT Presentation
October 23, 2001. Outline Metrics. 2. Outline Metrics. Outline Metrics are valuable in measuring the intricacy and

Presentation Transcript

Slide 1

´╗┐Outline Metrics CS 406 Software Engineering I Fall 2001 Aditya P. Mathur Last refresh: October 23, 2001

Slide 2

Design Metrics Design Metrics are helpful in measuring the many-sided quality and "goodness" of an outline. An expansive number measurements have been proposed for OO plans . Some of these have been approved tentatively, others are minor recommendations or have gotten practically no approval . Plan Metrics

Slide 3

Effort Assumption : The exertion in building up a class is dictated by the quantity of techniques. Subsequently the general many-sided quality of a class can be measured as a component of the unpredictability of its strategies. Proposition: Weighted Methods per class (WMC) Design Metrics

Slide 4

Let signify the multifaceted nature of strategy WMC Let class C have strategies M1, M2, .....Mn. How to quantify WMC? Plan Metrics

Slide 5

WMC: approval Most classes have a tendency to have few strategies, are basic, and give some particular deliberation and operations. WMC metric has a sensible relationship with blame inclination of a class. Outline Metrics

Slide 6

Depth of legacy tree Depth of a class in a class chain of importance decides potential for re-utilize. More profound classes have higher potential for re-utilize. Legacy builds coupling. Changing classes gets to be distinctly harder. Profundity of Inheritance (DIT) of class C is the length of the most brief way from the base of the legacy tree to C. On account of numerous legacy DIT is the most extreme length of the way from the root to C. Outline Metrics

Slide 7

DIT assessment Basili et al. study,1995. Chidamber and Kemerer contemplate, 1994. Most classes have a tendency to be near the root. Most extreme DIT esteem observed to be 10. Most classes have DIT=0. DIT is critical in foreseeing blunder inclination of a class. Higher DIT prompts to higher error-inclination. Outline Metrics

Slide 8

Number of kids (NOC) NOC is the quantity of quick subclasses of C. Higher estimations of NOC recommend reuse of the definitions in the super-class in a bigger number of subclasses. Higher NOC proposes the degree of impact of a class on different components of a plan. Higher impact requests higher nature of that class. Plan Metrics

Slide 9

Validation of NOC Classes for the most part have a little NOC esteem. Greater part have NOC=0. Bigger NOC esteem is related with lower likelihood of identifying deficiencies in that class. Outline Metrics

Slide 10

Coupling between classes (CBC) Class C1 is coupled to class C2 if no less than one technique for C1 utilizes a strategy or a case variable of C2. Coupling is generally simple to recognize however frequently pointers may make it troublesome. CBC of C=total number of different classes to which C is coupled. Plan Metrics

Slide 11

Validation of CBC Most classes are independent and have CBC=0. Interface classes have a tendency to have higher CBC values. CBC is huge in foreseeing deficiency inclination of classes. Plan Metrics

Slide 12

Response for a class (RFC) Response set of class C is the aggregate number of strategies that can be conjured when a message is sent to a protest of C. This incorporates all techniques for C and any strategies executed outside of C accordingly of this message. RFC of class C is the cardinality of the reaction set of C. Take note of that notwithstanding when CBC=1 RFC might be high. This shows the "volume" of association is high. Outline Metrics

Slide 13

Validation of RFC Most classes have a tendency to conjure few strategies (low RFC values). Classes for interface objects have a tendency to have bigger RFC values. RFC is extremely noteworthy in anticipating the blame inclination of a class. Plan Metrics

Slide 14

Lack of attachment in strategies (LCOM) [1] Let I1 and I2 signify sets of occurrence factors got to by techniques M1 and M2, separately, in class C. M1 and M2 are viewed as comparative, or durable, if I1 and I2 are not disjoint. Give Q a chance to be the arrangement of all firm strategy sets. Give P a chance to be the arrangement of all non-durable strategy sets. LCOM=|P| - |Q| if |P| > |Q|, 0 generally. Outline Metrics

Slide 15

LCOM [2] A bigger number of firm matches suggests littler LCOM. A high estimation of LCOM recommends that a class is attempting to bolster different deliberations. Maybe the class should be parceled into littler and more firm classes. LCOM is not observed to be extremely critical in foreseeing deficiency inclination. Plan Metrics

Slide 16

Summary What are OO measurements? Measurements for many-sided quality, coupling, and union Design Metrics

SPONSORS