F. Lanubile and G. Visaggio: "Evaluating Predictive Quality Models Derived from Software Measures: Lessons Learned". In Journal of Systems and Software, 38:225-234, 1997.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....that are likely to be faulty, for example by optimally allocating testing resources [22] or by redesigning components. Early identification of faulty components is commonly achieved through a binary quality model that classifies components into either a faulty or not faulty category [7][38]. These binary classification models almost always utilize product metrics as predictor variables. The theoretical basis for developing such binary classifiers has been articulated in [9] There it is hypothesized that the structural properties of a software component (such as its coupling) ....

....of nearest neighbors increases. However, it is more difficult to draw conclusions from these tables on which Although the use of principal components analysis was a possibility for reducing the number of variables, previous work does indicate that this may not improve the accuracy of classifiers [38]. particular combination of parameters produces the best predictive performance since the differences in the J coefficient are not marked. Equal Weight (standardization method) Distance Measure Z score Mean Absolute Median Absolute Euclidean 0.172 0.204 0.208 0.176 0.176 Table 3: ....

[Article contains additional citation context not shown here]

F. Lanubile and G. Visaggio: "Evaluating Predictive Quality Models Derived from Software Measures: Lessons Learned". In Journal of Systems and Software, 38:225-234, 1997.

....the consequences may be expensive field failures or costly defect correction later in the life cycle. 3.2.2. 3 Common Measures of Accuracy Common measures of accuracy for binary classifiers that have been applied in software engineering are the proportion correct value, reported in for example [1][41][56] also called correctness in [51] 52] completeness, for example [9] 51] which is equal to the senstivity) correctness which is the true positive rate [9] 12] 13] 14] also called consistency in [51] 52] the Kappa coefficient [12] 14] Type I and Type II misclassifications ....

F. Lanubile and G. Visaggio: "Evaluating Predictive Quality Models Derived from Software Measures: Lessons Learned". In Journal of Systems and Software, 38:225-234, 1997.

....problem. This is achieved through a quality model that classifies components into either a high or low risk category. The definition of a high risk component varies depending on the context of the study. For example, a high risk component is one that contains any faults found during testing [14][74], one that contains any faults found during operation [71] or one that is costly to correct after an error has been found [3] 13] 1] The identification of high risk components allows an organization to take mitigating actions, such as focus defect detection activities on high risk components, ....

F. Lanubile and G. Visaggio: "Evaluating Predictive Quality Models Derived from Software Measures: Lessons Learned". In Journal of Systems and Software, 38:225-234, 1997.

....programming guidelines, and to make system level predictions. These are described further below. 3.1 Identifying Risky Components The definition of a high risk component varies depending on the context. For example, a high risk component is one that contains any faults found during testing [11][82], one that contains any faults found during operation [75] or one that is costly to correct after a fault has been found [1] 5] 12] Recent evidence suggests that most faults are found in only a few of a system s components [41] 67] 91] 95] If these few components can be identified early, then ....

....is high risk or not is achieved through a quality model. A quality model is a quantitative model that can be used to: Predict which components will be high risk. For example, some quality models make binary predictions as to whether a component is faulty or not faulty [11] 30] 31] 35] 75][82]. Rank components by their risk proneness (in whatever way risk is defined) For instance, there have been studies that predict the number of faults in individual components (e.g. 72] and that produce point estimates of maintenance effort (e.g. 66] 84] These estimates can be used for ....

[Article contains additional citation context not shown here]

F. Lanubile and G. Visaggio, "Evaluating Predictive Quality Models Derived from Software Measures: Lessons Learned," Journal of Systems and Software, vol. 38, pp. 225-234, 1997.

....Below we discuss different approaches that can be used for evaluating prediction accuracy. 3.2.3. 1 Common Measures of Accuracy Common measures of accuracy for binary classifiers that have been applied in software engineering are the proportion correct value, reported in for example [1][45][63] also called correctness in [57] 58] completeness, for example [9] 57] which is equal to the senstivity) correctness which is the true positive V7 02 11 99 11 rate [9] 14] 12] 13] also called consistency in [57] 58] the Kappa coefficient [12] 14] Type I and Type II misclassifications ....

F. Lanubile and G. Visaggio: "Evaluating Predictive Quality Models Derived from Software Measures: Lessons Learned". In Journal of Systems and Software, 38:225-234, 1997.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC