Curriculum, Archetypes, Policies, Interventions & Research Environment
This repository is the public research portfolio for the CAPIRE research programme. It presents the conceptual architecture, manuscript map, methodological contributions, and public-safe outputs of CAPIRE without including raw institutional data, private manuscript drafts, or the private analytical codebase.
The private analytical repository is not public. It contains student-level data governed by institutional privacy requirements and is maintained separately under controlled access.
Student attrition and delayed graduation in higher education are rarely explained by individual-level factors alone. They emerge from the interaction of institutional structures, socioeconomic context, and temporal dynamics that conventional predictive models fail to capture adequately.
CAPIRE was designed to address this gap by treating student trajectories as longitudinal, structurally constrained processes — not as static classification targets.
- Structural bottlenecks in curriculum design as measured on directed acyclic prerequisite graphs.
- Temporal dynamics of academic progression, with explicit attention to observation time and data leakage.
- Student trajectory archetypes discovered through Topological Data Analysis (TDA), not imposed a priori.
- The causal impact of macro-level shocks (economic crises, institutional disruptions) on individual trajectories.
- Social synchrony patterns and peer-effect structures within administrative enrolment groups.
- Policy-level interventions simulated through agent-based modelling calibrated on empirical archetypes.
CAPIRE integrates five methodological layers:
- Leakage-aware data layer — strict chronological feature construction preventing temporal data leakage.
- Curriculum graph analytics — prerequisite DAG analysis to quantify structural friction.
- Archetype discovery — TDA (Mapper, UMAP, DBSCAN) for student population segmentation.
- Causal and macro-shock modelling — structural causal models and event-aligned time-series analysis.
- Agent-based simulation and intervention lab — empirically calibrated ABM for policy scenario testing.
See docs/architecture.md for the detailed technical description.
The CAPIRE research programme spans multiple manuscript families:
| Family | Topic | Public Status |
|---|---|---|
| Leakage-aware data layer | Multilevel trajectory modelling framework | Public preprint |
| Curriculum graph | Structural bottleneck analysis | Under preparation |
| Conceptual framework | Psychohistory, TDA, and ABM integration | Internal / submitted |
| Regularity Trap | Curriculum-structural archetype | Under preparation |
| Structural Amplifier | Archetype-level amplification effects | Under preparation |
| Macro-shocks | Inflation and strike impacts on trajectories | Under preparation |
| Social Synchrony | Administrative network and peer-effect analysis | Under preparation |
| Intervention Lab / JASSS | Agent-based policy simulation | Internal / submitted |
| Promotion Wall | Late-stage structural barriers | Under preparation |
| AI-era curriculum redesign | Civil engineering curriculum reconceptualisation | Public — Zenodo |
Full details: docs/manuscript_map.md and manuscripts/public_index.md.
Confirmed public artefacts:
- Preprint: Paz, H. R. (2025). A Leakage-Aware Data Layer for Student Analytics: The CAPIRE Framework for Multilevel Trajectory Modeling. Research Square. https://doi.org/10.21203/rs.3.rs-8118343/v1
- Preprint: Paz, H. R. (2026). Toward a New Curricular Architecture for Civil Engineering in the Age of Artificial Intelligence. Zenodo. https://doi.org/10.5281/zenodo.19591967
- Preprint: Paz, H. R. (2026). AlphaPath: Toward a Search-Based Architecture for Professional Knowledge Navigation (v1.0.0). Zenodo. https://doi.org/10.5281/zenodo.19552912
See docs/public_outputs.md for the full list.
Raw institutional data are not included in this repository and are not publicly available.
- Student-level academic records are governed by institutional privacy requirements and are never publicly released.
- Processed row-level features derived from institutional data are also non-public.
- Public reproducibility relies on synthetic or aggregate artefacts only.
- Any future public release of data will use synthetically generated datasets that preserve statistical structure without mapping to real individuals.
See docs/data_governance.md for the full data governance policy.
CAPIRE operates under a tiered reproducibility model:
- Conceptual reproducibility: The methodological architecture, algorithms, and analytical logic are fully documented.
- Synthetic-data reproducibility: Selected workflows are designed to run with synthetic inputs (in preparation).
- Aggregate reproducibility: Publication-safe figures and tables are traceable to their generating scripts.
- Institutional-data reproducibility: Full empirical reproduction requires authorised access to private institutional data and is not publicly available.
See docs/reproducibility_statement.md.
This is a documentation-first public research portfolio. It does not contain the private analytical codebase.
CAPIRE-research-portfolio/
├── README.md This file
├── CITATION.cff Citation metadata
├── LICENSE CC BY 4.0
├── .gitignore
├── docs/ Public documentation
│ ├── overview.md
│ ├── research_programme.md
│ ├── architecture.md
│ ├── methods_summary.md
│ ├── manuscript_map.md
│ ├── reproducibility_statement.md
│ ├── data_governance.md
│ ├── public_outputs.md
│ ├── citation_guide.md
│ └── roadmap.md
├── figures/ Public-safe figures (populated as approved)
│ ├── architecture/
│ ├── manuscript_maps/
│ └── public_safe/
├── manuscripts/ Public manuscript index and metadata
│ ├── public_index.md
│ └── metadata/
├── references/ Bibliographic references
│ ├── references.bib
│ └── apa_references.md
└── portfolio/ Research portfolio summaries
├── one_page_summary.md
├── research_line_summary.md
└── public_profile.md
If you refer to the CAPIRE framework or this portfolio, please use:
Paz, H. R. (2025). A Leakage-Aware Data Layer for Student Analytics:
The CAPIRE Framework for Multilevel Trajectory Modeling.
Research Square. https://doi.org/10.21203/rs.3.rs-8118343/v1
See CITATION.cff and docs/citation_guide.md for full citation options.
Hugo Roger Paz Doctoral Researcher ORCID: 0000-0003-1237-7983
This repository is under active development as a public research portfolio. The analytical codebase and private manuscript artefacts remain in a private repository. Public reproducibility artefacts, synthetic datasets, and additional figures will be added as they are reviewed and approved for public release.
For academic enquiries, please contact via ORCID profile or institutional channels.
This repository documents the CAPIRE research programme for public academic communication. It does not contain raw institutional data, student-level records, private manuscript drafts, or private codebase components. The empirical findings described in associated manuscripts are specific to the institutional context and datasets analysed; generalisation requires appropriate context mapping and recalibration.