#### DMCA

## Empirical studies of quality models in object-oriented systems (2002)

### Cached

### Download Links

- [sdmetrics.com]
- [squall.sce.carleton.ca]
- [www.idi.ntnu.no]
- DBLP

### Other Repositories/Bibliography

Venue: | Advances in Computers |

Citations: | 42 - 5 self |

### Citations

5808 | Classification and Regression Trees - Breiman, Friedman, et al. - 1984 |

2168 |
Applied logistic regression.
- Hosmer, Lemeshow
- 2000
(Show Context)
Citation Context ...reen out measures that are not significantly related to the DV and not likely to be significant predictors in multivariate models. Only measures that are significant at significance level, say α=0.25 =-=[HL89]-=-, should be considered for the subsequent multivariate analysis. Note that some IV may be significantly related to the DV for various reasons. It may capture a causal relationship or be the result of ... |

1665 |
Signal detection theory and psychophysics.
- Green, Swets
- 1966
(Show Context)
Citation Context ...near model” in [BMW01]) π Standard measures of the goodness of fit used in the context of logistic regression models are sensitivity, specificity, and the area under the receiver-operator-curve (ROC) =-=[GS74]-=-. Sensitivity is the fraction of observed positive outcome cases that are correctly classified (i.e., the fraction of faulty classes correctly classified faultprone, which similar to completeness as d... |

1346 |
Regression models for categorical and limited dependent variables.
- Long
- 1997
(Show Context)
Citation Context ...xplain the variance of the DV and ultimately obtain accurate predictions. To measure the prediction accuracy, different modeling techniques (e.g., OLS [Lew80], logistic regression, Poisson regression =-=[Lon97]-=-) have specific measures of goodness of fit of the model. 5. Prediction model evaluation In order to get an estimate of the predictive power of the multivariate prediction models that is more realisti... |

1216 |
A complexity measure.
- McCabe
- 1976
(Show Context)
Citation Context ...e complexities ci were intentionally left undefined. Two versions of WMC were suggested and are frequently used: • In [LH93] and [LHKS95], ci is defined as McCabe’s cyclomatic complexity of method Mi =-=[McC76]-=-. • In [BBM96], each ci is set to one. In other words, this version of WMC counts the (non-inherited) methods of the class. The Appendix provides short definitions for all measures mentioned in this c... |

1156 |
Cross-validation choice and assessment of statistical predictions
- Stone
- 1974
(Show Context)
Citation Context ...tions, with some degree of success [BMW01][BW01]. • Evaluating prediction models. From the studies that perform multivariate analysis, only half of these studies perform some kind of cross validation =-=[Sto74]-=-, where the prediction performance of the multivariate prediction model in a relevant application context is investigated. The other studies only provide a measure of the goodness of fit of the predic... |

1080 | A metrics suite for object oriented design”,
- R, Kemerer
- 1994
(Show Context)
Citation Context ...epth. CLD of a class is the maximum number of levels in the hierarchy that are below the class. [TSM95] NOC Number of children. The number of classes that directly inherit from a given class. [CK91], =-=[CK94]-=- NOP Number of parents. The number of classes that a given class directly inherits from. [LC94], [LK94] NOD Number of descendents. The number of classes that directly or indirectly inherit from a clas... |

673 | Multivariate adaptive regression splines
- Friedman
- 1991
(Show Context)
Citation Context ...linear relationships or interactions. In addition, because of the lack of supporting theory, it is often difficult to know what to search for. Exploratory techniques, such as regression trees or MARS =-=[Fri91]-=-, have been used in some studies to determine non-linearities and variable interactions, with some degree of success [BMW01][BW01]. • Evaluating prediction models. From the studies that perform multiv... |

567 |
Regression Diagnostics - Identifying Influential Data and Sources of Collinearity.
- Belsley, Kuh, et al.
- 2004
(Show Context)
Citation Context ...el. Overfitting a model increases the standard error of the model’s prediction, making the model more dependent on the data set it is based on and thus less generalizable. • Reduces multicollinearity =-=[BKW80]-=-, i.e., independent variables which are highly correlated. High multicollinearity results in large standard errors for regression coefficient estimates and may affect the predictive power of the model... |

433 | A validation of object-oriented design metrics as quality indicators.
- Basili, Briand, et al.
- 1996
(Show Context)
Citation Context ... ci were intentionally left undefined. Two versions of WMC were suggested and are frequently used: • In [LH93] and [LHKS95], ci is defined as McCabe’s cyclomatic complexity of method Mi [McC76]. • In =-=[BBM96]-=-, each ci is set to one. In other words, this version of WMC counts the (non-inherited) methods of the class. The Appendix provides short definitions for all measures mentioned in this chapter. 2.3 Su... |

347 |
Simultaneous Statistical Inference
- Miller
- 1966
(Show Context)
Citation Context ...of measures, based on a clear rationale, is indispensable to maintain the complexity of the data analysis within reasonable limits and lower the chances of finding significant relationships by chance =-=[Mil81]-=-. However, in the early stage of investigation, it is common for studies to investigate large numbers of alternatives, as they tend to be exploratory. • Building prediction models. Only about half of ... |

269 |
Object-oriented metrics that predict maintainability”
- Li, Henry
- 1995
(Show Context)
Citation Context ...E95], [CN00], [LC94]). Besides empirical studies, the literature is concerned with the following topics: 5. Definition of new sets of measures (e.g., [BDM97], [BK95], [CK91], [CK94], [Hen96], [HM95], =-=[LH93]-=-, [LK94], [LLWW95], [Li98]) 6. Definition of measurement frameworks for one or more structural properties, which provide guidelines how these properties can, in principle, be measured [EKS94], [HM95],... |

257 | A unified Framework for coupling Measurement in ObjectOriented Systems,"
- Briand, Daley, et al.
- 1999
(Show Context)
Citation Context ...[LLWW95], [Li98]) 6. Definition of measurement frameworks for one or more structural properties, which provide guidelines how these properties can, in principle, be measured [EKS94], [HM95], [BDW98], =-=[BDW99]-=-. 7. Criticism/theoretical analysis of existing measures and measurement frameworks; in particular, there is an interest in defining, for measures of various structural properties, necessary mathemati... |

205 | A unified framework for cohesion measurement in object-oriented systems”
- Briand, Daly, et al.
- 1998
(Show Context)
Citation Context ... [LK94], [LLWW95], [Li98]) 6. Definition of measurement frameworks for one or more structural properties, which provide guidelines how these properties can, in principle, be measured [EKS94], [HM95], =-=[BDW98]-=-, [BDW99]. 7. Criticism/theoretical analysis of existing measures and measurement frameworks; in particular, there is an interest in defining, for measures of various structural properties, necessary ... |

204 | Property-Based Software Engineering Measurement,"
- Briand, Morasca, et al.
- 1996
(Show Context)
Citation Context ...defining, for measures of various structural properties, necessary mathematical properties these measures must possess in order for them to be valid measures of the properties they purport to measure =-=[BMB96]-=-, [KPF95], [Whi97], [Zus98]. Our discussions in this article will focus on categories 1) and 2), with a strong emphasis on the former as these studies are by far the most numerous. 2.2 Measurement In ... |

193 |
Object-Oriented Systems Analysis: Modeling the World
- Shlaer, Mellor
- 1988
(Show Context)
Citation Context ... two or more systems, e.g., [AGE95], [Bar98], [CK94], [SC93]. For instance, [SC93] develop two versions of a brewery 2scontrol system to identical specifications, one following a data-driven approach =-=[SM88]-=-, the other following a responsibility-driven approach [WWW90]. They apply the set of design measures by Chidamber and Kemerer [CK91] to the two resulting systems. They find the system resulting from ... |

190 | Towards a metrics suite for object oriented design
- Chidamber, Kemerer
- 1991
(Show Context)
Citation Context ... to identical specifications, one following a data-driven approach [SM88], the other following a responsibility-driven approach [WWW90]. They apply the set of design measures by Chidamber and Kemerer =-=[CK91]-=- to the two resulting systems. They find the system resulting from the responsibility-driven approach to display more desirable structural properties. They conclude the responsibility-driven to be mor... |

185 | Exploring the relationship between design measures and software quality in object-oriented systems”,
- Briand, Wüst, et al.
- 2000
(Show Context)
Citation Context ...to balance the number of actually faulty and predicted fault-prone classes. Plotting the correctness and completeness curves as a function of the selected threshold p0 is also a good, common practice =-=[BWDP00]-=-, as shown in Figure 1. As an example, we show here the fault-proneness classification results from a model (“Linear” logistic regression Model) built in [BMW01]: Table 4: Fault-proneness classificati... |

184 | N.: Towards a Framework for Software Measurement Validation
- Kitchenham, Pfleeger, et al.
(Show Context)
Citation Context ... for measures of various structural properties, necessary mathematical properties these measures must possess in order for them to be valid measures of the properties they purport to measure [BMB96], =-=[KPF95]-=-, [Whi97], [Zus98]. Our discussions in this article will focus on categories 1) and 2), with a strong emphasis on the former as these studies are by far the most numerous. 2.2 Measurement In this sect... |

157 | Cohesion and reuse in an objectoriented system
- Bieman, Kang
- 1995
(Show Context)
Citation Context ...and performing factor analysis (e.g., [AGE95], [CN00], [LC94]). Besides empirical studies, the literature is concerned with the following topics: 5. Definition of new sets of measures (e.g., [BDM97], =-=[BK95]-=-, [CK91], [CK94], [Hen96], [HM95], [LH93], [LK94], [LLWW95], [Li98]) 6. Definition of measurement frameworks for one or more structural properties, which provide guidelines how these properties can, i... |

139 |
Logistic Regression Diagnostics
- Pregibon
- 1981
(Show Context)
Citation Context ...ystematically tested. For many regression techniques, specific diagnostics to automatically identify outliers were proposed, e.g., Cooks Distance for OLS [BP95], Pregibon beta for logistic regression =-=[Pre81]-=-. 3.4 Prediction Model Construction Multivariate regression is performed to build prediction models of the DV. This analysis is conducted to determine how well we can predict the DV, when the design m... |

132 | An Investigation into Coupling Measures for C
- Briand, Devanbu, et al.
- 1997
(Show Context)
Citation Context ...elations and performing factor analysis (e.g., [AGE95], [CN00], [LC94]). Besides empirical studies, the literature is concerned with the following topics: 5. Definition of new sets of measures (e.g., =-=[BDM97]-=-, [BK95], [CK91], [CK94], [Hen96], [HM95], [LH93], [LK94], [LLWW95], [Li98]) 6. Definition of measurement frameworks for one or more structural properties, which provide guidelines how these propertie... |

108 |
The Confounding Effect of Class Size on the Validity of ObjectOriented Metrics,"
- Emam, Benlarbi, et al.
- 2001
(Show Context)
Citation Context ...- 100% versions - CK94, LH93 4sReference Dependent Variable Independent Data Set Univariate Multivariate Model Evaluation Variable analysis Analysis BEGR00 fault-proneness (from C&K, without LCOM see =-=[EBGR01]-=- and LR R-Sq; LL Ratio Test field faults) [EBGR99] to compare models BM99 fault-proneness (from Suite of LALO, see [BWL01] LR LR goodness of fit (chifield faults) polymorphism measures by authors, C&K... |

104 | Managerial use of metrics for object-oriented software: An exploratory analysis
- Chidamber, Darcy, et al.
- 1998
(Show Context)
Citation Context ...ased coupling 6. Inh. depth below 5. Inh. depth above 4. Export coupling 2. Size 3. norm. cohesion How built 1. Import coupling Source Modeling Technique, IV stepwise heuristic LOC 2 7 adj. R 2 = 75% =-=[CDK98]-=- OLS, Productivity LOC, HICBO 4 HILCOM 7 adj. R 2 = 82% HILCOM HICBO 3 7 adj. R 2 = 60% HILCOM HICBO 2 adj. R 2 2 = 60% OLS, Rework effort OLS Design effort DIT NOC 10 adj. R 2 WMC NOM Size1 Size2 MPC... |

103 | Measuring Coupling and Cohesion in Object-Oriented Systems
- Hitz, Montazeri
- 1995
(Show Context)
Citation Context ....g., [AGE95], [CN00], [LC94]). Besides empirical studies, the literature is concerned with the following topics: 5. Definition of new sets of measures (e.g., [BDM97], [BK95], [CK91], [CK94], [Hen96], =-=[HM95]-=-, [LH93], [LK94], [LLWW95], [Li98]) 6. Definition of measurement frameworks for one or more structural properties, which provide guidelines how these properties can, in principle, be measured [EKS94],... |

81 |
Designing Object-Oriented
- Wirfs-Brock, Wiener
- 1990
(Show Context)
Citation Context ... For instance, [SC93] develop two versions of a brewery 2scontrol system to identical specifications, one following a data-driven approach [SM88], the other following a responsibility-driven approach =-=[WWW90]-=-. They apply the set of design measures by Chidamber and Kemerer [CK91] to the two resulting systems. They find the system resulting from the responsibility-driven approach to display more desirable s... |

70 | The Prediction of Faulty Classes Using Object-Oriented Design Metrics,”
- Emam, Melo
- 1999
(Show Context)
Citation Context ...s, C++, commercial LR R-square field faults) NMO,SIX, avg. telecom frame work, #parameters, LOC 174 classes EBML00 fault-proneness (from Stmts, Mthds, Attrs; see [EBGR01], LR (Threshold field faults) =-=[EMM01]-=-, plus commercial telecom app, C++, 85 classes, models) EBGR99 fault-proneness (from C&K, C-Food see EBML00 LR LR R-square, leavefield faults) one-out CV, receiver-operator curve (ROC) EMM01 fault-pro... |

66 | Assessing the applicability of faultproneness models across object-oriented software projects
- Briand, Melo, et al.
(Show Context)
Citation Context ...hat to search for. Exploratory techniques, such as regression trees or MARS [Fri91], have been used in some studies to determine non-linearities and variable interactions, with some degree of success =-=[BMW01]-=-[BW01]. • Evaluating prediction models. From the studies that perform multivariate analysis, only half of these studies perform some kind of cross validation [Sto74], where the prediction performance ... |

65 |
A Framework of Software Measurement. Walter de Gruyter
- Zuse
- 1998
(Show Context)
Citation Context ...arious structural properties, necessary mathematical properties these measures must possess in order for them to be valid measures of the properties they purport to measure [BMB96], [KPF95], [Whi97], =-=[Zus98]-=-. Our discussions in this article will focus on categories 1) and 2), with a strong emphasis on the former as these studies are by far the most numerous. 2.2 Measurement In this section, we provide so... |

62 |
Object-Oriented Metrics
- Henderson-Sellers
- 1996
(Show Context)
Citation Context ...alysis (e.g., [AGE95], [CN00], [LC94]). Besides empirical studies, the literature is concerned with the following topics: 5. Definition of new sets of measures (e.g., [BDM97], [BK95], [CK91], [CK94], =-=[Hen96]-=-, [HM95], [LH93], [LK94], [LLWW95], [Li98]) 6. Definition of measurement frameworks for one or more structural properties, which provide guidelines how these properties can, in principle, be measured ... |

60 | Validation of the coupling dependency metric as a predictor of run-time failures and maintenance measures
- Binkley, Schach
- 1998
(Show Context)
Citation Context ...al dependency (extent to which a program relies on its internal organization remaining unchanged, 3) data integrity dependency (vulnerability of data elements in one module to change by other module) =-=[BS98]-=- Fan-In Count of modules (classes) that call a given class, plus the number of global data elements [BS98] Fan-Out Count of modules (classes) called by a given module plus the number of global data el... |

52 | Replicated Case Studies for Investigating Quality Factors
- Briand, Wüst, et al.
(Show Context)
Citation Context ...ariable analysis Analysis BEGR00 fault-proneness (from C&K, without LCOM see [EBGR01] and LR R-Sq; LL Ratio Test field faults) [EBGR99] to compare models BM99 fault-proneness (from Suite of LALO, see =-=[BWL01]-=- LR LR goodness of fit (chifield faults) polymorphism measures by authors, C&K, Part of C-FOOD square) BS96? EO: agreement with CDM, CBO, DAC, 8 system specs, 2-3 ad hoc, count experts’ preference NCC... |

50 | Coupling and cohesion in object-oriented systems
- Eder, Kappel, et al.
- 1994
(Show Context)
Citation Context ..., [HM95], [LH93], [LK94], [LLWW95], [Li98]) 6. Definition of measurement frameworks for one or more structural properties, which provide guidelines how these properties can, in principle, be measured =-=[EKS94]-=-, [HM95], [BDW98], [BDW99]. 7. Criticism/theoretical analysis of existing measures and measurement frameworks; in particular, there is an interest in defining, for measures of various structural prope... |

46 | Toward the design quality evaluation of object-oriented software systems
- Abreu, Goulão, et al.
- 1995
(Show Context)
Citation Context ... a set of design measures to one or more systems; with a discussion of the obtained distributions of the measures within one system, or a comparison of distributions across two or more systems, e.g., =-=[AGE95]-=-, [Bar98], [CK94], [SC93]. For instance, [SC93] develop two versions of a brewery 2scontrol system to identical specifications, one following a data-driven approach [SM88], the other following a respo... |

41 | Experimental Assessment of the Effect of Inheritance on the Maintainability of Object-Oriented Systems
- Harrison, Counsell, et al.
- 2000
(Show Context)
Citation Context ...maintainable than the equivalent flat versions. In a third experimental using a larger system no difference was found between a system using 5 levels of inheritance and an equivalent flat version. In =-=[HCN00]-=-, students performed modifications on two pairs of functionally equivalent systems that differ in the use of inheritance (no inheritance, three/five levels of inheritance). The correctness and complet... |

36 |
Another metric suite for object-oriented programming
- Li
- 1998
(Show Context)
Citation Context ...ides empirical studies, the literature is concerned with the following topics: 5. Definition of new sets of measures (e.g., [BDM97], [BK95], [CK91], [CK94], [Hen96], [HM95], [LH93], [LK94], [LLWW95], =-=[Li98]-=-) 6. Definition of measurement frameworks for one or more structural properties, which provide guidelines how these properties can, in principle, be measured [EKS94], [HM95], [BDW98], [BDW99]. 7. Crit... |

36 |
Eliciting and Analyzing Expert Judgement: A Practical Guide
- Meyer, Booker
- 1991
(Show Context)
Citation Context ... carefully planned to provide meaningful results and the procedures used must be properly reported. Though this is outside of the scope of this article, some abundant literature exists on the subject =-=[MB91]-=-. An interesting question that, to our knowledge, has not been investigated in depth to date is whether structural measures can perform as well as or better than experts in predicting quality attribut... |

34 |
The Object-Oriented Brewery: A Comparison of Two Object-Oriented Development Methods
- Sharble, Cohen
- 1993
(Show Context)
Citation Context ...to one or more systems; with a discussion of the obtained distributions of the measures within one system, or a comparison of distributions across two or more systems, e.g., [AGE95], [Bar98], [CK94], =-=[SC93]-=-. For instance, [SC93] develop two versions of a brewery 2scontrol system to identical specifications, one following a data-driven approach [SM88], the other following a responsibility-driven approach... |

33 | Some misconceptions about lines of code,” - Rosenberg - 1997 |

31 |
A New Metric for Object-Oriented Design
- Chen, Lum
- 1993
(Show Context)
Citation Context ...(as visible in high R-squared>99.9%). • The models in [BM99] and [NQ98] contain covariates with p-values above 0.5. Such covariates can be removed from the model without significant loss of fit. • In =-=[CL93]-=-, a full model with all candidate measures is fitted. From this model, not significant covariates were removed, and predicted values were calculated from the remaining significant covariates, but reta... |

31 |
Effort Estimation and Prediction for Object-Oriented Systems
- Nesi, Querci
- 1998
(Show Context)
Citation Context ...models. Including them would bias the results: • The models in [AM96] based on 8 observations and 7 covariates suffer from overfitting (as visible in high R-squared>99.9%). • The models in [BM99] and =-=[NQ98]-=- contain covariates with p-values above 0.5. Such covariates can be removed from the model without significant loss of fit. • In [CL93], a full model with all candidate measures is fitted. From this m... |

23 |
Coupling Metrics for ObjectOriented Design
- Harrison, Counsell, et al.
- 1998
(Show Context)
Citation Context ...includes inheritance-based coupling (coupling between classes related via inheritance). NAS Number of associations - count of the number of association lines emanating from a class in an OMT diagram. =-=[HCN98]-=- COF Coupling Factor - percentage of pairs of classes that are coupled. [AM96] CDM Coupling Dependency Metric. Sum of 1) referential dependency (extent to which a program relies on its declaration dep... |

23 |
Object-Oriented Design Measurement
- Whitmire
- 1997
(Show Context)
Citation Context ...ures of various structural properties, necessary mathematical properties these measures must possess in order for them to be valid measures of the properties they purport to measure [BMB96], [KPF95], =-=[Whi97]-=-, [Zus98]. Our discussions in this article will focus on categories 1) and 2), with a strong emphasis on the former as these studies are by far the most numerous. 2.2 Measurement In this section, we p... |

22 |
Evaluating the Impact
- F, Melo
- 1996
(Show Context)
Citation Context ... ordinary R-square density, rework effort systems, university least- squares (R-sq); (system wide) setting, from students, 4 to 15 KLOC regression (LS) BBM96 fault-proneness C&K [CK94]; Code UMD, see =-=[AM96]-=- Logistic Logistic Contingency Table, (faults from Metrics Nesting Regression (LR) regression (LR) Correctness/ acceptance testing) Level, FunctDef, FunctCall Completeness BWDP00 fault-proneness ~50 m... |

22 |
Principal Components Analysis. Sage University
- Dunteman
- 1989
(Show Context)
Citation Context ...y distributions of the IVs. This will help to explain some of the results observed in subsequent steps and is also crucial to explain differences across studies. 2. Principal component analysis (PCA) =-=[Dun89]-=- In the investigation of measures of structural properties, it is common to have much collinearity between measures capturing similar underlying phenomena. PCA is a standard technique to determine the... |

20 |
Polymorphism measures for early risk prediction
- Benlarbi, Melo
- 1999
(Show Context)
Citation Context ...K, commercial, 40 ness/Completeness, CFOOD. KLOC 10- CV BMW01 fault-proneness (field Suite of XPOSE: 144 classes, LR LR, MARS R-sq, Correct faults) polymorphism commercial ness/Completeness, measures =-=[BM99]-=-, 10- CV, cross-sys C&K, Part of Ctem validation, costFOOD benefit model BW01 development effort ~50 measures LIOO, university negative binomial Poisson 10-CV, ARE, MRE including C&K, C- setting, publ... |

19 | Multi-method research: An empirical investigation of objectoriented technology
- Wood
- 1999
(Show Context)
Citation Context ...the shallow systems, while taking the same time as the deep versions. Modification tasks took shorter for the deep versions, but were carried with lower correctness than on the shallow versions. • In =-=[WDMR99]-=-, students performed modifications on three pairs of functionally equivalent systems. The time they required for the modification was measured. In two experimental runs, systems with 3 levels of inher... |

16 | Use of Factor Analysis to Develop OOP Software Complexity Metrics
- Lake, Cook
- 1994
(Show Context)
Citation Context ...et of design measures to one or more systems and investigate relationships between these design measures, by investigating pairwise correlations and performing factor analysis (e.g., [AGE95], [CN00], =-=[LC94]-=-). Besides empirical studies, the literature is concerned with the following topics: 5. Definition of new sets of measures (e.g., [BDM97], [BK95], [CK91], [CK94], [Hen96], [HM95], [LH93], [LK94], [LLW... |

15 |
An Evaluation of Code Metrics for ObjectOriented Programs
- Harrison, Samaraweera, et al.
- 1996
(Show Context)
Citation Context ...arman rho HCN98 fault density, EO: NMI, CBO, NAS (197 cls.); Subjective LEDA lib. (97 cls.); complexity 3 sets of student (understandability) on systems, 113, 172, a 1-5 scale 317 cls. resp; syst. in =-=[HSDL96]-=- HN96 EO: Subjective C&K, LOC C++ system, 13 cls., Spearman, complexity [HC98] 500LOC, poss. Kendall, Pear Comm. son’s r HSDL96 EO: Subjective LOC, lib/non-lib C++ system, 109 Spearman, complexity (HC... |

13 |
A Software Complexity Metric for C++,
- Lake, Cook
- 1992
(Show Context)
Citation Context ...ple systems were too small to allow for any statistical testing of significant differences. The remaining studies all focus on the impact of inheritance on understandability and maintainability. • In =-=[LC92]-=-, students answered comprehension questions and performed debugging and modification tasks on functionally equivalent systems with deep and shallow inheritance hierarchies. The time required to perfor... |

12 | Reliability and maintainability related software coupling metricsin C++ programs
- Rajaraman, Lyu
- 1992
(Show Context)
Citation Context ...lationships between inherited and new/redefined methods (should be similar or the same as AMMIC). [TKC99] CC Class coupling - number of method couplings, i.e., variable references and/or method calls =-=[RL92]-=- AMC Average method coupling - CC divided by number of methods in the class. [RL92]sName Definition Source CBO Coupling between object classes. According to the definition of this measure, a class is ... |

10 | An experimental evaluation of quality guidelines on the maintainability of object-oriented design documents
- Briand, Bunse, et al.
- 1997
(Show Context)
Citation Context ...d. The OO system displaying good object-oriented design principles was shown to be better understandable and faster to modify than the version violating the design principles. A refined experiment in =-=[BBD97]-=- confirmed these results. Note however that no statement can be made whether any of the design principles contributed more or less to the differences in understandability/maintainability. • Though not... |

9 |
Measuring the Coupling and Cohesion of an Object-Oriented
- Lee, Liang, et al.
- 1995
(Show Context)
Citation Context ...C94]). Besides empirical studies, the literature is concerned with the following topics: 5. Definition of new sets of measures (e.g., [BDM97], [BK95], [CK91], [CK94], [Hen96], [HM95], [LH93], [LK94], =-=[LLWW95]-=-, [Li98]) 6. Definition of measurement frameworks for one or more structural properties, which provide guidelines how these properties can, in principle, be measured [EKS94], [HM95], [BDW98], [BDW99].... |

8 |
Using Coupling Measurement for Impact Analysis
- Briand, Wust, et al.
- 1999
(Show Context)
Citation Context ... is easier to account for but leads to other practical problems. If systems become the unit of analysis then it becomes difficult to obtain enough data to perform multivariate analysis. • Two studies =-=[BWL99]-=-, [WH98] used the likelihood or number of ripple effects to other classes when a change is performed to a class. The goal was to provide a model to support impact analysis. These studies are not descr... |

7 | The impact of design properties on development cost in Object-Oriented systems.
- Briand, Wust
- 2001
(Show Context)
Citation Context ...search for. Exploratory techniques, such as regression trees or MARS [Fri91], have been used in some studies to determine non-linearities and variable interactions, with some degree of success [BMW01]=-=[BW01]-=-. • Evaluating prediction models. From the studies that perform multivariate analysis, only half of these studies perform some kind of cross validation [Sto74], where the prediction performance of the... |

7 |
Use of friends in C++ software: An empirical investigation
- Counsell, Newson
(Show Context)
Citation Context ...pply a set of design measures to one or more systems and investigate relationships between these design measures, by investigating pairwise correlations and performing factor analysis (e.g., [AGE95], =-=[CN00]-=-, [LC94]). Besides empirical studies, the literature is concerned with the following topics: 5. Definition of new sets of measures (e.g., [BDM97], [BK95], [CK91], [CK94], [Hen96], [HM95], [LH93], [LK9... |

5 | Coupling Measures and Change Ripples - Wilkie, Kitchenham - 1998 |

4 | Impediments to the Effective Use of Metrics within the Object-Oriented Paradigm - Binkley, Schach - 1996 |

4 | and Jürgen Wüst. “Integrating scenario-based and measurementbased software product assessment
- Briand
(Show Context)
Citation Context ...ersions produced on a regular basis. • Design measurement can be combined with more qualitative analyses of software systems to make the results of a qualitative analysis more objective. For example, =-=[BW01b]-=- describes a study aimed at assessing the modifiability and reusability of a software system. To this end, a set of change scenarios 36sthat the system is likely to undergo in the future was identifie... |

4 | Estimation of Effort and Complexity: An Object-oriented Case Study - Misic, Tesic - 1998 |

4 |
The Hybrid CART-Logit Model
- Steinberg, Cardell
- 1999
(Show Context)
Citation Context ...s and automatically searches for interactions. Both techniques can be combined with traditional regression modeling [BW01]. Hybrid Models with Regression Trees Adapting some of the recommendations in =-=[SC99]-=-, traditional regression analysis and regression trees can be combined into a hybrid model as follows: Run regression trees analysis, with some restrictions on the minimum number of observations in ea... |

3 |
C.: An Experimental Comparison of the Maintainability
- Briand, Bunse, et al.
- 1997
(Show Context)
Citation Context ... or a proper backward elimination procedure could have been used. 4.2 Controlled Experiments In the following, we briefly report on the qualitative results established in controlled experiments. • In =-=[BBDD97]-=-, students answered comprehension questions and performed change impact analysis on functionally equivalent systems with differing adherence to design principles (including low coupling, high cohesion... |

3 |
Measuring Complexity
- Wilkie
- 1998
(Show Context)
Citation Context ...r to account for but leads to other practical problems. If systems become the unit of analysis then it becomes difficult to obtain enough data to perform multivariate analysis. • Two studies [BWL99], =-=[WH98]-=- used the likelihood or number of ripple effects to other classes when a change is performed to a class. The goal was to provide a model to support impact analysis. These studies are not described in ... |

2 |
Validating Object- Oriented Design Metrics on a commercial Java application
- Glasberg, Emam, et al.
- 2000
(Show Context)
Citation Context ...or size does not systematically “invalidate” measures in their studies. For example, OCMIC is found insignificant after controlling for size in [EBGR99], but significant after controlling for size in =-=[GEMM00]-=-. Furthermore, in the next section, where multivariate models are reported to predict fault-proneness, a number of studies have shown that models that are based on both coupling and size perform signi... |

2 | The role of inheritance in the maintainability of object-oriented systems
- Harrison, Counsell
- 1998
(Show Context)
Citation Context .... (97 cls.); complexity 3 sets of student (understandability) on systems, 113, 172, a 1-5 scale 317 cls. resp; syst. in [HSDL96] HN96 EO: Subjective C&K, LOC C++ system, 13 cls., Spearman, complexity =-=[HC98]-=- 500LOC, poss. Kendall, Pear Comm. son’s r HSDL96 EO: Subjective LOC, lib/non-lib C++ system, 109 Spearman, complexity (HC98); functions called, functions, 2.5KLOC. Kendall, Pear #faults in testing, d... |

2 |
Applied Regression: An Introduction”, Sage publications
- Lewis-Beck
- 1980
(Show Context)
Citation Context ...tes in a multivariate model, in order to better explain the variance of the DV and ultimately obtain accurate predictions. To measure the prediction accuracy, different modeling techniques (e.g., OLS =-=[Lew80]-=-, logistic regression, Poisson regression [Lon97]) have specific measures of goodness of fit of the model. 5. Prediction model evaluation In order to get an estimate of the predictive power of the mul... |

2 |
An Empirical Study on Object-Oriented
- Tang, Kao, et al.
- 1999
(Show Context)
Citation Context ...ere c is using another class. FMMEC EC: Export coupling: count interactions where class d is the used class. DMMEC OMMEC IC Inheritance Coupling: Number of parent classes to which a class is coupled. =-=[TKC99]-=- CBM Coupling between methods: Number of function dependency relationships between inherited and new/redefined methods (should be similar or the same as AMMIC). [TKC99] CC Class coupling - number of m... |

1 | A new reusability metrics for object-oriented measures - Barnard - 1998 |

1 |
An Empirical Investigation of an Object
- Cartwright, Shepperd
(Show Context)
Citation Context ... 3 R 2 =0.4355 Leaveone-out specificity 0.83 AROC: 0.78, Proportion correct: 0.77 Betweensystem validation AROC: 0.85 92% completeness, [GEMM00] OCMIC NMImp OCMEC DIT, DIT 2 5 R 2 = 0.42 Leaveone-out =-=[CS00]-=- OLS EVNTS 1 adj. R 2 = 87.2% EVNTS INHTS 2 adj. R 2 = 89% [BWDP00] LR forward stepwise NM, 3 R heuristic NMpub, NumPar 2 =0.139, 67% completeness, 60% correctness NIHICP, FMMEC NOP CLD 7 R RFC, NMI R... |

1 |
Research Design”, Newbury Park
- Spector
- 1981
(Show Context)
Citation Context ...involved. The rightmost column indicates what experimental design was employed and the analysis techniques used to test the research hypotheses. For an introduction to experimental designs, see e.g., =-=[Spe81]-=-. Table 2: Overview of controlled experiments Referenc Dependent variable Independent variables Systems/Subjects Exp. Design/ e An. Technique Bar98 reusability (subjective C&K, LOC, Methods, 2 systems... |