# An Exploratory Big Data Approach to Understanding Commitment in Projects

**Authors:** Narasimha Rao Vajjhala (University of New York Tirana, Tirana, Albania) — ORCID 0000-0002-8260-2392; Kenneth David Strang (W3 Research, New York, USA; and University of the Cumberlands, Williamsburg, KY, USA) — ORCID 0000-0002-4333-4399
**Type:** Conference paper
**Source:** WorldCIST 2024 (World Conference on Information Systems and Technologies), Lecture Notes in Networks and Systems, vol. 989, pp. 66–75, Springer Nature Switzerland AG
**Published:** 2024
**DOI:** https://doi.org/10.1007/978-3-031-60227-6_6
**Canonical page:** https://www.narasimharao.net/research/machine-learning-project-manager-organizational-commitment-big-data/
**Indexing:** Scopus · Web of Science
**Keywords:** Project Management; Organizational Commitment; Machine Learning; Big Data Analysis; Talent Retention; Project Failure Rates; Data-Driven Management; Predictive Modeling; Stakeholder Implications; Program Management; Performance Indicators

## Summary

**Question.** Can project performance indicators in retrospective big data serve as a reliable, bias-free gauge of project manager organizational commitment, and which attributes of successful projects predict PM commitment tenure?

**Finding.** Analyzing 439 successful U.S. (primarily military) IT projects from 2015–2023 with supervised machine learning, the linear regression model best predicted PM commitment tenure (r² = 0.238, MAE = 5.31), ahead of random forest (r² = 0.108) and SVM (r² = 0.047). Line of business, PM experience, network collaboration tools and cross-industry team membership were the strongest features, supporting the proposition that higher PM commitment is associated with successful projects.

**Why it matters.** The study offers a more objective, big data-driven way to understand project manager commitment based on actual behavioral evidence rather than self-reported surveys, which suffer from bias and small effect sizes. Its insights are aimed at global stakeholders in projects and programs concerned with retaining experienced talent and reducing high project failure rates.

## Key findings

1. The study analyzed 439 successful U.S. IT projects (2015–2023), mainly from U.S. military divisions, drawn from a declassified government big data repository of more than 500 projects and 1,230 features.
2. Project managers in the sample of 439 IT projects had an average age of 40.2 years, 54.3% were male, nearly a third were PMI certified, and average PM commitment tenure was 11.08 years (range 1–25).
3. Among three machine learning models predicting project manager commitment tenure, linear regression performed best (r² = 0.238, MAE = 5.31, RMSE = 6.40), compared with random forest (r² = 0.108) and support vector machine (r² = 0.047).
4. Line of business was the most influential feature for PM commitment tenure (relative relief coefficient 0.360), with ocean/water-based projects more successful than air space and land-based projects.
5. PM experience (0.293), network collaboration using internet tools (0.277), cross-industry team membership (0.270) and contracted rather than in-house PMs (0.222) were the next strongest predictors of PM commitment tenure.
6. Multi-dimensional scaling revealed distinct clusters of features related to PM commitment tenure, but weak silhouette scores led the authors to shift to machine learning regression.
7. The authors conclude that experience, network collaboration, cross-industry collaboration and sponsor co-location support the proposition that increased PM organizational commitment is associated with successful projects.

## Study at a glance

| Item | Detail |
|---|---|
| Research question | RQ1: Does PM organizational commitment in successful projects correlate with other project attributes? RQ2: Which attributes of successful projects predict PM commitment tenure? |
| Design | Exploratory, pragmatic machine learning study of retrospective secondary big data |
| Data | Declassified U.S. government repository of IT projects (2015–2023), mainly from U.S. military divisions: 1,230 features on 500+ projects, cleaned to 439 successful projects |
| Methods | Radial analysis, multi-dimensional scaling, and supervised ML regression (linear regression, random forest, SVM) evaluated with MSE, RMSE, MAE and r²; relative relief feature ranking |
| Main result | Linear regression explained 23.8% of variance in PM commitment tenure; line of business, experience, network collaboration and cross-industry membership ranked highest |
| Implication | Behavioral big data can objectively gauge PM commitment and inform talent retention in projects |

## Abstract

This study addresses the twin challenges of talent retention and high project failure rates (40–70%) by harnessing machine learning (ML) techniques to analyze retrospective big data. The study’s objective was to ascertain whether project performance indicators can be a reliable gauge of project manager (PM) organizational commitment. This approach sidesteps the inherent bias and small effect sizes associated with survey self-report responses. Our innovative methodology leverages secondary big data, transforming the values into structured features that predict PM organizational commitment. This study proposes a novel conceptual framework, focusing on actual behavioral evidence rather than traditional, self-reported attitudes to assess the fuzzy predictors of organizational commitment. Among the three developed ML models, one demonstrated a significant 24% effect size, uncovering key features correlating PM tenure and organizational commitment with success. The insights gained from this research have broad implications for global stakeholders in projects and programs, offering a more objective and big data-driven understanding of PM commitment.

## Key terms

- **Organizational commitment:** An employee's psychological attachment and loyalty to their organization; in this study, PM tenure was the critical measure of commitment.
- **Machine learning (ML):** A subset of artificial intelligence that lets computer systems perform tasks through algorithms and statistical models without explicit instructions, relying on pattern recognition and inference.
- **Coefficient of determination (r²):** The proportion of variance in the target variable explained by a model, used here as the effect size of each ML regression model.

## Limitations

- The data were retrospective secondary data, and it was ambiguous whether all IT projects involved military personnel or subcontractors.
- The study needs to be replicated using different samples and with triangulation of methods beyond machine learning.

## How to cite

Vajjhala, N. R., & Strang, K. D. (2024). An Exploratory Big Data Approach to Understanding Commitment in Projects. In Á. Rocha (Ed.), WorldCIST 2024 (World Conference on Information Systems and Technologies) (pp. 66–75). Springer Nature Switzerland AG. https://doi.org/10.1007/978-3-031-60227-6_6

```bibtex
@inproceedings{vajjhala2024machine,
  title = {An Exploratory Big Data Approach to Understanding Commitment in Projects},
  author = {Vajjhala, Narasimha Rao and Strang, Kenneth David},
  booktitle = {WorldCIST 2024 (World Conference on Information Systems and Technologies)},
  series = {Lecture Notes in Networks and Systems},
  editor = {Á. Rocha},
  pages = {66--75},
  year = {2024},
  publisher = {Springer Nature Switzerland AG},
  doi = {10.1007/978-3-031-60227-6_6},
  url = {https://doi.org/10.1007/978-3-031-60227-6_6}
}
```