KPI DB used in a cost-driven framework for Technical Debt remediation (in Agile software teams – 7 years of evidence base)

DATASET STRUCTURE AND CONTENTS

The JSON dataset is organized as a nested time-series KPI model per project, capturing the weekly evolution of performance indicators. The hierarchical structure is as follows:

·        Project – Top-level entries represent individual software projects (each project is identified by an anonymized code). Each project entry includes high-level information such as an overall KPI score or health index for that project.

·        Area – Within each project, data are grouped by major area of performance or concern (e.g., “Client Satisfaction”, “Delivery”, “Quality Assurance”). Each area has an aggregated areaStatus score reflecting the project’s status in that domain.

·        Category – Under each area, specific categories correspond to individual KPIs or sub-metrics (e.g., “Perceived Quality”, “Execution Speed”, “On Budget”, “Test Coverage”). Each category entry contains a current status numerical score for that KPI, along with explanatory and planning fields.

·        Weekly Timeline – For each category, a weekly time-series record captures the KPI values over the project timeline (organized by week). This temporal data illustrates how each KPI evolved week by week, enabling trend analysis over the project’s lifecycle.

Each category entry provides a numeric KPI value (score) and is accompanied by textual annotations: a “why” field that explains the context or rationale behind the current score, and an “actions” field noting any remedial steps or management decisions related to that KPI. The data model follows a consistent time-series schema: weekly snapshots of KPI scores are indexed in the JSON structure to facilitate chronological analysis. Higher-level composite scores (such as area-level or project overall scores) are computed from the underlying category metrics using defined aggregation logic – for example, some aggregate indices are calculated using geometric mean formulas – to ensure that no single outlier KPI dominates the overall rating. This algorithmic scoring approach (including use of geomean aggregation and structured categorical scoring rubrics) is documented in the associated research and is implicitly reflected in the dataset, providing analytical traceability from raw weekly metrics up to summary scores.

RESEARCH CONTEXT AND USAGE

This KPI dataset underpins two (possible 3 in the next future) associated peer-reviewed publications that focus on technical debt estimation, scenario-based analysis, and KPI modeling in software engineering projects. The data were pivotal in developing and validating new methods for estimating technical debt using scenario-based development metrics and for modeling project health through composite KPIs. The JSON dataset alone is sufficient to reproduce and validate the quantitative claims and methodological findings presented in those studies. In particular, all statistical results, trend analyses, and model evaluations reported in the articles can be derived directly from the anonymized data provided here, without reliance on the private source documents. By archiving this dataset on a public repository, we ensure that other researchers and practitioners can verify the published findings and reuse the data for further analysis. In summary, the dataset offers a rigorously sanitized yet rich account of project performance trajectories, suitable for independent validation of the associated studies’ conclusions and for broader investigations into KPI-based project assessment and technical debt modeling in software development.