11-Software metrics

Quantitative Software Metrics

• Code Metrics
  • Static
  • Dynamic
• Management Metrics
  • Cost
  • Duration, Time
  • Staffing
• Testing Metrics
• Productivity Metrics
• Design Metrics
• Maintainability Metrics

Software Quality Management

• Ensures required quality levels within budget & time constraints.
• Quality software must have minimal defects and meet standards (maintainability, reliability, etc.).
• Requires a framework at both organizational and project levels.

Software Measurement

• Derives numeric values for attributes of software products/processes for objective comparisons.
• Limited systematic use in many organizations; few established standards exist.

Software Metrics

• Any measurement related to software system, process, or documentation, e.g.:
  • Lines of code
  • Fog readability index
  • Person-days to develop component.
• Used for predicting product attributes and controlling the software process.

Types of Process Metrics

• Time Metrics: Duration to complete processes.
• Resource Metrics: Efforts in person-days, costs, etc.
• Event Occurrences: Track defects, requirements changes, etc.

Measurement Objectives

• Understand development processes:
  • Cost-effectiveness of inspections
  • Effect of pair programming on productivity.
  • Project completion rates (on time/budget).

Predictor and Control Measurements

• Control Metrics: Assess product.
• Predictor Metrics: Inform management decisions.

Use of Measurements

• Assign values to system quality attributes (e.g., LOC, complexity).
• Identify sub-standard components by measuring characteristics.

Metrics Assumptions

• Accurate measurable software properties exist.
• Relationships between measurable and desired attributes need validation.

Product Metrics

• Dynamic Metrics: Measure during program execution.
• Static Metrics: Measure system representations.
• Quality metrics should predict product quality.

Dynamic vs. Static Metrics

• Dynamic Metrics: Directly relate to quality attributes (e.g., system response time).
• Static Metrics: Indirectly relate to quality; e.g., complexity and maintainability.

Software Component Analysis

• Analyze components using various metrics; compare values to identify issues.
• Establish thresholds and monitor anomalous measurements.

Measurement Ambiguity

• The context of collected data is crucial; incorrect interpretations can lead to misleading insights.

Software Context

• Measured processes/products are affected by changing business environments, requiring careful analysis of results.

Measurement Surprises

• Enhanced quality may lead to increased help desk calls due to user expectations shifting.

Problems in the Software Industry

• ROI quantification of metrics programs is challenging.
• Lack of standardization and definition in processes persists across companies.

Empirical Software Engineering

• Bases itself on data collection and experiments to validate software engineering methods and techniques.

Software Analytics

• Focuses on understanding and improving software systems through data analysis and insights.

Analytics Tools Requirements

• User-friendly, quick outputs, comprehensive measurement capabilities, and interactive exploratory analysis.

Status of Software Analytics

• Still developing, with uncertainties regarding effectiveness; dependent on collected data metrics family.

Metrics Areas Empirically Researched

• Covers productivity, design, testing, maintainability, management, and code metrics.

Deciding What to Measure

• Identify customer goals and relevant metrics to assess success accordingly.

Measurement Caveats

• Software data is often imprecise; manage impact of measurement on processes and balance costs against benefits.