Metrics As Scores Dataset: Metrics and Domains From the Qualitas.class Corpus

This dataset was created by extracting software metrics data from the Qualitas.class corpus (Terra et al. 2013; Tempero et al. 2010). Therefore, the principal quantity type is Metric, and the context is given by a system’s Domain (e.g., "Game", "Middleware", etc.). Some metrics were obtained on program-level, while others are package- or method-level metrics. Most of the metrics in the corpus are of discrete/integral nature. The corpus holds 23 types of pre-computed software metrics for a total of 111 systems which are spread across eleven different domains.

This dataset has the following discrete Features (Metrics):

CA: Afferent Coupling

CE: Efferent Coupling

DIT: Depth of Inheritance Tree

MLOC: Method Lines of Code

NBD: Nested Block Depth

NOC: Number of Classes

NOF: Number of Attributes

NOI: Number of Interfaces

NOM: Number of Methods

NOP: Number of Packages

NORM: Number of Overridden Methods

NSC: Number of Children

NSF: Number of Static Attributes

NSM: Number of Static Methods

PAR: Number of Parameters

TLOC: Total Lines of Code

VG: McCabe Cyclomatic Complexity

WMC: Weighted Methods per Class

The following features are continuous:

LCOM: Lack of Cohesion in Methods

RMA: Abstractness

RMD: Normalized Distance

RMI: Instability

SIX: Specialization Index

It has a total of 11 Groups (Domains): 3D; Graphics; Media, Databases, Diagrams; Visualiz., Games, IDE, Middleware, Parsers; Generators, Progr. Language, SDK, Testing, and Tool.