# Machine Learning and Data Science for ESG Compliance Measurement in Financial Software Engineering Projects: A Single-Case Socio-Technical Systems Analysis

**Authors:** Kenneth David Strang (University of the Cumberlands, Williamsburg, KY, USA; W3-Research) — ORCID 0000-0002-4333-4399; Narasimha Rao Vajjhala (Department of Computer Science, American University in Bulgaria, Blagoevgrad, Bulgaria) — ORCID 0000-0002-8260-2392
**Type:** Journal article (open access)
**Source:** Systems, 14(8), Article 990
**Published:** 2026-08-14
**DOI:** https://doi.org/10.3390/systems14080990
**Canonical page:** https://www.narasimharao.net/research/machine-learning-esg-compliance-financial-software-projects/
**Indexing:** Scopus Q1 (CiteScore 5.4) · Web of Science SSCI Q1 · Impact Factor 3.8 (JCR 2025)

## Summary

Strang and Vajjhala (2026) applied an open-source data science and machine learning workflow to 207 anonymized archival project records from one firm, in a single-case study of Environmental, Social, and Governance (ESG) compliance measurement in financial software engineering projects. Stakeholder-rated social and governance factors were almost perfectly correlated with the overall project score (r = +0.995 and +0.966), while the environmental factor was unrelated to it. Exploratory classifiers performed weakly (kNN AUC = 0.497; SVM accuracy 61.8%), and the correlations are read as a property of the firm’s evaluation system.

**Practical implication.** For organizations measuring project-level ESG compliance, measurement from stakeholder-rated records reflects the firm’s rating practice as much as the algorithm, on single-case evidence from one firm.

- **The Problem:** Organizations increasingly want to measure Environmental, Social, and Governance (ESG) compliance at the level of individual software engineering projects, but it is unclear what machine learning can learn from the evaluation records firms actually produce.
- **The Methodology:** Single-case socio-technical systems study of archival project records (N = 207) using an open-source data science workflow: correlation analysis, principal component analysis, common-method-bias diagnostics, and kNN and SVM classification.
- **The Core Finding:** In 207 financial software project records, stakeholder-rated social and governance factors tracked the overall project score almost perfectly (r = +0.995 and +0.966) while the environmental factor was unrelated to it, and machine learning classifiers performed weakly (kNN AUC = 0.497; SVM accuracy 61.8%). ESG measurement from such records reflects the firm’s rating practice as much as the algorithm — a socio-technical systems result.
- **The Citation:** Strang, K. D., & Vajjhala, N. R. (2026). Machine Learning and Data Science for ESG Compliance Measurement in Financial Software Engineering Projects: A Single-Case Socio-Technical Systems Analysis. Systems, 14(8), Article 990. https://doi.org/10.3390/systems14080990

## Key findings

- **Finding 1.** The study applied an open-source data science and machine learning workflow to 207 anonymized archival project records from a financial software engineering firm.
- **Finding 2.** Stakeholder-rated social and governance factors had near-unity correlations with the overall project score (r = +0.995 and r = +0.966, p < 0.001).
- **Finding 3.** The environmental factor was unrelated to the overall project score.
- **Finding 4.** Post-hoc diagnostics — a seven-component principal-component structure, Harman’s single-factor screen, near-zero same-source correlations, and marker-variable partial correlations — bound but cannot eliminate method-based explanations, so the correlations are read as a property of the firm’s evaluation system.
- **Finding 5.** Exploratory classifiers performed weakly: k-nearest neighbors was at chance (AUC = 0.497) and the support vector machine reached 61.8% accuracy, modestly above the no-information baseline.
- **Finding 6.** The gap between near-unity correlations and weak classification is explained by different feature sets: the near-redundant social and governance ratings were withheld from the classifiers.

Source: Strang & Vajjhala (2026), Systems, 14(8), Article 990. https://doi.org/10.3390/systems14080990

## The study in detail

**Question.** Can an open-source data science and machine learning workflow measure Environmental, Social, and Governance (ESG) compliance in financial software engineering projects — and what does a firm’s own evaluation practice allow such models to learn?

**Finding.** In 207 anonymized project records from one financial firm, the stakeholder-rated social and governance factors were almost perfectly correlated with the overall project score (r = +0.995 and +0.966), while the environmental factor was unrelated to it. Machine learning classifiers performed weakly: kNN was at chance (AUC = 0.497) and SVM reached 61.8% accuracy, only modestly above baseline.

**Why it matters.** The study shows that ESG measurement is shaped as much by the human rating process and organizational templates as by the algorithm — a socio-technical systems view. The near-unity correlations describe how the firm scores projects rather than distinct ESG constructs, and the weak classifiers show the limits of learning from such records. It offers a proof of concept and a research agenda for AI-enabled, project-level ESG measurement.

## Methods and data

| Item | Detail |
|---|---|
| Research question | How can AI and data science measure ESG compliance at the level of individual software engineering projects? |
| Design | Single-case study treating one firm’s project evaluation practice as a socio-technical system |
| Data | 207 anonymized archival project records from a financial software engineering firm |
| Methods | Open-source data science and machine learning workflow; correlation analysis; principal component analysis and method-bias diagnostics; kNN and SVM classifiers |
| Main result | Social (r = +0.995) and governance (r = +0.966) ratings track the overall project score; environmental factor unrelated; kNN AUC = 0.497, SVM accuracy 61.8% |
| Implication | Proof of concept and a structured agenda for AI-enabled, project-level ESG measurement |

## Definitions and key terms

- **ESG compliance:** The degree to which an organization or project meets Environmental, Social, and Governance criteria set by regulators, investors, or internal policy. (as used in Strang & Vajjhala (2026))
- **Socio-technical system:** A system in which people, organizational rules, and technology jointly determine outcomes, so that none can be understood in isolation. (as used in Strang & Vajjhala (2026))
- **Common method bias:** Inflated correlations that arise because several variables were measured by the same source or instrument rather than because the constructs are truly related. (as used in Strang & Vajjhala (2026))

## Abstract

The integration of artificial intelligence (AI) and data science into organizational evaluation practices creates socio-technical systems in which human rating behavior, organizational templates, regulatory pressure, and analytical algorithms jointly determine what can be measured and learned. This single-case study examines the measurement of Environmental, Social, and Governance (ESG) compliance in financial software engineering projects, treating one firm’s project evaluation practice as a socio-technical system and applying an open-source data science and machine learning workflow to 207 anonymized archival project records. Correlation analysis revealed near-unity associations between the stakeholder-rated social and governance factors and the overall project score (r = +0.995 and +0.966, p < 0.001), while the environmental factor was unrelated to the score; post-hoc diagnostics (a seven-component principal-component structure, Harman’s screen, selective near-zero same-source correlations, and marker-variable partial correlations) bind, but cannot eliminate, method-based explanations, so the coefficients are interpreted as a descriptive property of the firm’s evaluation system rather than as estimates of relationships between validated, distinct constructs. Exploratory machine learning classifiers performed weakly—kNN at chance (AUC = 0.497) and SVM only modestly above the no-information baseline (accuracy 61.8%)—a result consistent with the constraints that the social subsystem imposes on the learnability of the records it generates, although technical factors, including the dichotomization of the target variable, the modest sample size, and model configuration, cannot be ruled out as contributing explanations; descriptive statistics are reported for all variables, and a diagnostic analysis reconciles the apparent divergence between the near-unity correlations and the weak classification performance by showing that the two rest on different feature sets, the near-redundant social and governance ratings having been withheld from the classifiers. The findings offer a proof of concept and a structured agenda for AI-enabled, project-level ESG measurement in socio-technical systems.

## Limitations and open questions

- Single-case design: one firm’s evaluation practice, so results are descriptive of that system.
- Classifier performance may also reflect dichotomization of the target variable, modest sample size, and model configuration.

### Open questions

- Can prospective, time-stamped data capture and penalized estimation separate genuine predictive signal from leakage in organizational ESG and risk records?

Open questions from the evidence overview: [Can ESG compliance be measured reliably? Evidence from ESG ratings, project-level instruments, machine learning and supply chain records](https://www.narasimharao.net/research/topics/esg-compliance-measurement-ratings-projects-supply-chains/)

## Practical implications

- Organizations that measure ESG compliance from stakeholder-rated project records are measuring something shaped as much by the human rating process and organizational templates as by the algorithm, as a single-case study of one firm shows.
- For organizations applying machine learning to evaluation records, weak classifiers on one firm’s records are consistent with limits on learning from such records, although dichotomization of the target variable, modest sample size and model configuration may also contribute.
- For organizations working with stakeholder-rated project evaluations, the near-unity correlations found in one firm (r = +0.995 and +0.966) describe how the firm scores projects rather than distinct ESG constructs.
- Organizations that want AI-enabled, project-level ESG measurement can draw on the study as a proof of concept and a research agenda rather than as a large-scale benchmark.

## When this research may be relevant

This article may be relevant to literature searches on machine learning for ESG measurement, sustainability compliance in software engineering or IT projects, ESG in financial services firms, socio-technical analysis of AI adoption, and the limits of learning from stakeholder-rated project evaluation data. It is a proof-of-concept and research-agenda paper rather than a large-scale benchmark.

## How to cite

**APA 7:** Strang, K. D., & Vajjhala, N. R. (2026). Machine Learning and Data Science for ESG Compliance Measurement in Financial Software Engineering Projects: A Single-Case Socio-Technical Systems Analysis. Systems, 14(8), Article 990. https://doi.org/10.3390/systems14080990

**MLA 9:** Strang, Kenneth David, and Narasimha Rao Vajjhala. “Machine Learning and Data Science for ESG Compliance Measurement in Financial Software Engineering Projects: A Single-Case Socio-Technical Systems Analysis.” Systems, vol. 14, no. 8, 2026, article 990. https://doi.org/10.3390/systems14080990.

**IEEE:** K. D. Strang and N. R. Vajjhala, “Machine Learning and Data Science for ESG Compliance Measurement in Financial Software Engineering Projects: A Single-Case Socio-Technical Systems Analysis,” Systems, vol. 14, no. 8, Art. no. 990, Aug. 2026, doi: 10.3390/systems14080990.

```bibtex
@article{strang2026machine,
  title = {Machine Learning and Data Science for ESG Compliance Measurement in Financial Software Engineering Projects: A Single-Case Socio-Technical Systems Analysis},
  author = {Strang, Kenneth David and Vajjhala, Narasimha Rao},
  journal = {Systems},
  volume = {14},
  number = {8},
  pages = {990},
  year = {2026},
  publisher = {MDPI},
  doi = {10.3390/systems14080990},
  url = {https://doi.org/10.3390/systems14080990}
}
```

## Related publications

- Salahudeen et al. (2026). [Talking Face Generation in Socio-Technical Systems: A Systematic Review of Deep Learning Architectures, Deployment Contexts, and Governance Frameworks](https://www.narasimharao.net/research/talking-face-generation-socio-technical-systems-systematic-review/). https://doi.org/10.3390/systems14101198. Across 92 primary studies (2016–2025), Diffusion Transformers and Gaussian Splatting account for about 36% of studies dated 2024 or later while GAN use has fallen below 3%; training data are demographically skewed and under-documented, and the dominant metrics (PSNR, SSIM, SyncNet) do not measure what deployment requires.
- Strang & Vajjhala (2024). [Evaluating the Anti-Corruption Factor in Environmental, Social, and Governance Indices by Sampling Large Financial Asset Management Firms](https://www.narasimharao.net/research/esg-ratings-anti-corruption-asset-management-firms/). https://doi.org/10.3390/su162310240. For 14 asset managers, each with USD 1 trillion or more in assets under management, no ESG score from MSCI, Sustainalytics or CSRhub was significantly correlated with misconduct or arbitration case counts, and MSCI and Sustainalytics scores for the same firms were uncorrelated (r = −0.069).
- Strang & Vajjhala (2026). [Regulatory Volatility in Digital Supply Chains: An Information Systems Analytics Study of Decision-Maker Risk Perceptions and Sustainability-Related Outcomes](https://www.narasimharao.net/research/regulatory-volatility-digital-supply-chains-esg-risk/). https://doi.org/10.3390/su18189359. Perceived regulatory-volatility risks do not form a single construct (KMO = 0.470), and labor and environmental risks were rated lowest of all (medians 0 and 1 on 0–5 scales) during a period of record forced-labor enforcement.
- Strang & Vajjhala (2026). [Verifying SDG ESG Compliance in Manufacturing Industry Projects by Surveying Sponsors](https://www.narasimharao.net/research/esg-compliance-manufacturing-projects-survey-instrument/). https://doi.org/10.3390/info17040311. A survey of 2,231 North American manufacturing project sponsors produced a validated seven-item, two-factor instrument — ESG planning and ESG monitoring and controlling — with excellent fit (CFI = 0.99, TLI = 0.98, RMSEA = 0.052, SRMR < 0.035).
- Strang & Vajjhala (2025). [Exploring project manager commitment using machine learning on fuzzy big data](https://www.narasimharao.net/research/project-manager-commitment-machine-learning-big-data/). https://doi.org/10.1504/IJPOM.2025.146727. Across 439 successful US government IT projects with 1,230 features, linear regression outperformed random forest and SVM in predicting project manager commitment tenure.

## Discussed in

- Research synthesis: [Machine learning & data science](https://www.narasimharao.net/research/focus-areas/machine-learning-data-science/)
- Research synthesis: [ESG & sustainability measurement](https://www.narasimharao.net/research/focus-areas/esg-sustainability-measurement/)
- Research synthesis: [Sustainable software engineering](https://www.narasimharao.net/research/focus-areas/sustainable-software-engineering/)
- Evidence overview: [Can ESG compliance be measured reliably? Evidence from ESG ratings, project-level instruments, machine learning and supply chain records](https://www.narasimharao.net/research/topics/esg-compliance-measurement-ratings-projects-supply-chains/)