Back

Data Set

File Change History

Format: csv

Fields:

• change_id: String used to identify a change throughout the entire dataset. It is in the following format <FILE_NUMBER>_change<NUMBER>. Changes are numbered in reverse chronological order, i.e. change1 always represents the most recent modification to a file.
• commit_hash: Git identification hash used to reference a commit.
• current_path: Path where the file currently resides.
• old_path: If the file was moved in that change (or created in that change, when value is None), it is the previous path where the file used to be located. If the file wasn’t moved, it equals current_path.

Observations: N/A

Download Link

File Changes with Metrics

Format: csv

Fields:

• change_id: String used to identify a change throughout the entire dataset. It is in the following format <FILE_NUMBER>_change<NUMBER>. Changes are numbered in reverse chronological order, i.e. change1 always represents the most recent modification to a file.
• commit_hash: Git identification hash used to reference a commit.
• current_path: Path where the file currently resides.
• old_path: If the file was moved in that change (or created in that change, when value is None), it is the previous path where the file used to be located. If the file wasn’t moved, it equals current_path.
• PercentLackOfCohesion: LCOM (Software Metric)
• CountClassCoupled: CBO (Software Metric)
• MaxNesting: MaxNEST (Software Metric)
• SumCyclomatic: Sum Cyclomatic Complexity (Software Metric)
• SumCyclomaticModified: Sum Modified Cyclomatic Complexity (Software Metric)
• SumCyclomaticStrict: Sum Strict Cyclomatic Complexity (Software Metric)
• SumEssential: Sum Essential Complexity (Software Metric)
• AvgCyclomatic: Average Cyclomatic Complexity (Software Metric)
• AvgCyclomaticModified: Average Modified Cyclomatic Complexity (Software Metric)
• AvgCyclomaticStrict: Average Strict Cyclomatic Complexity (Software Metric)
• AvgEssential: Average Essential Complexity (Software Metric)
• CountDeclMethod: Local Methods (Software Metric)
• MaxCyclomatic: Max Cyclomatic Complexity (Software Metric)
• MaxCyclomaticModified: Max ModifiedCyclomatic Complexity (Software Metric)
• MaxCyclomaticStrict: Max Strict Cyclomatic Complexity (Software Metric)
• MaxEssential: Max Essential Complexity (Software Metric)
• CountLine: Physical Lines (Software Metric)
• CountLineBlank: Blank LOC (Software Metric)
• CountLineCode: LOC (Software Metric)
• CountLineCodeDecl: Declarative Lines of Code (Software Metric)
• CountLineCodeExe: Executable Lines of Code (Software Metric)
• CountLineComment: CLOC (Software Metric)
• CountSemicolon: Semicolons (Software Metric)
• CountStmt: Statements (Software Metric)
• CountStmtDecl: Declarative Statements (Software Metric)
• CountStmtExe: Executable Statements (Software Metric)
• RatioCommentToCode: Comment to Code Ratio (Software Metric)
• CountDeclInstanceVariable: NIV (Software Metric)
• CountDeclInstanceMethod: NIM (Software Metric)
• CountDeclMethodAll: RFC - Response for Class (Software Metric)
• CountDeclMethodDefault: Local Default Visibility Methods (Software Metric)
• AvgLine: Average Number of Lines (Software Metric)
• AvgLineBlank: Average Number of Blank Lines (Software Metric)
• AvgLineCode: Average Number of Lines of Code (Software Metric)
• AvgLineComment: Average Number of Lines with Comments (Software Metric)
• CountDeclClassMethod: Class Methods (Software Metric)
• CountDeclClassVariable: Class Variables (Software Metric)
• CountDeclMethodPrivate: Private Methods (Software Metric)
• CountDeclMethodProtected: Protected Methods (Software Metric)
• CountDeclMethodPublic: Public Methods (Software Metric)
• CountClassBase: IFANIN (Software Metric)
• CountClassDerived: NOC (Software Metric)
• MaxInheritanceTree: DIT (Software Metric)
• TightClassCohesion: TCC (Software Metric)
• FANOUT: FANOUT (Software Metric)
• FANIN: FANIN (Software Metric)
• LCOM2: LCOM2 (Software Metric)
• LCOM3: LCOM 3 (Software Metric)
• ChangeSet: Change Set (Process-related metric)
• HunksCount: Hunks Count (Process-related metric)
• CodeChurn: Code Churn (Process-related metric)
• ContributorCount: Contributor Count (Developer-related metric)
• FileContributionPercentage: File Contribution Percentage (Developer-related metric)
• RefactoringCount: Total Number of Refactorings (Process-related metric)
• MoveRefactoringCount: Number of Move Refactorings (Process-related metric)
• ExtractionRefactoringCount: Number of Extraction Refactorings (Process-related metric)
• HierarchicalRefactoringCount: Number of Hierarchical Refactorings (Process-related metric)
• RenameRefactoringCount: Number of Rename Refactorings (Process-related metric)
• num_pulls_classes: Number of Associated Pull-Requests (Developer-related metric)
• words_in_discussion_by_class: Number of Words in Discussion (Developer-related metric)
• words_per_comment_by_class: Number of Words per Comment (Developer-related metric)

Observations: This file contains additional metrics which weren’t used in the analysis presented in the paper.

Download Link

Decay Score

Format: csv

Fields:

• file_id: Integer that identifies a file throughout the whole dataset.
• decay_score: Score, ranging from 0 to 1, that was calculated for each file. More details in step 5 of section III.B of the paper.

Observations: N/A

Download Link

Tagged Changes

Format: csv

Fields:

• change_id: String used to identify a change throughout the entire dataset. It is in the following format <FILE_NUMBER>_change<NUMBER>. Changes are numbered in reverse chronological order, i.e. change1 always represents the most recent modification to a file.
• tags: List of tags (divided by the | character) that were attributed to each change. More details in Step 7 of section III.B.

Observations: N/A

Download Link

Association Rules

Format: csv

Fields:

• Items: All tags present in the association rule.
• Antecedents: Antecedents for the association rule.
• Consequent: Consequents for the association rule.
• Support: Support metric.
• Confidence: Confidence metric.
• Lift: Lift metric.

Observations: Separated into two csvs for each project. They are named <PROJECT>_low for rules inferred from slightly-decayed classes and <PROJECT>_high for ones inferred from largely-decayed classes.

A third set of CSVs, containing rules from all classes is present in the second link.

Download Link

Download Link (All classes)

Summarized Rules

Format: csv

Fields:

• items: All tags present in the association rule.
• antecedents: Antecedents for the association rule.
• consequent: Consequents for the association rule.
• support_low: Support metric.
• support_high: Support metric.
• confidence_low: Confidence metric.
• confidence_high: Confidence metric.
• lift_low: Lift metric.
• lift_high: Lift metric.

Observations: Since these csvs aggregate the association rules from both rules, the support, confidence and lift metrics are split. They are named <METRIC>_low when inferred from slightly-decayed classes and <METRIC>_high when inferred from largely-decayed classes.

Download Link