The library
Data, evidence & assurance
What would make the picture trustworthy?
Match methods to the available information. Examine data maturity, project controls and what earlier prediction experiments could establish.
Examples, diagrams and reading sit together here. Dates and limitations belong to each piece; an earlier illustration is not a claim of current practice.
Earlier exploration · Information & methods
Which method fits the question and the available data?
Earlier thinking on matching methods to project information. Follow an illustrated Orange prediction workflow, ask what a synthetic delay forecast could know before work starts, or consider what a monthly portfolio review should decide and learn. Historical sources and later reconstructions are distinguished.
Interactive example
Discuss data maturity in context
Inspect maturity stages, dimensions and their connections to organisational capabilities.
What to try & what it leaves out
Try. Compare the question bank in graph and table views. Filter by capability or level and open the detailed questions to discuss what evidence an answer would need.
Limits. A supplied discussion framework, not a validated assessment instrument or an evidence-based score for your organisation.
Earlier example, repaired and reviewed 2 October 2026. This review checks the stated model and interactions; it does not establish usefulness in practice.
GitHub source (may require access)
Interactive example
Check the shape and meaning of project data
Edit a fictional project record and inspect checks on dates, costs, funding and provenance.
What to try & what it leaves out
Try. Complete a milestone early, then change a CAPEX funding currency and set tolerance to zero. Revalidate and compare the findings.
Limits. Illustrative checks, not official PDS validation or certification. Funding assumes whole-baseline coverage; provenance is declared, not independently verified.
Earlier example, repaired and reviewed 2 October 2026. This review checks the stated model and interactions; it does not establish usefulness in practice.
GitHub source (may require access)
2020 idea · synthetic reconstruction, 2026
What could an activity-delay forecast know in time?
Ask what a delay forecast could know before work starts. A seeded fictional schedule preserves the 2020 question while making pre-start inputs, held-out checks and simple baselines explicit.
Interactive example
Inspect a governance trail
Inspect a simulated governance trail, then check the records and evidence in a fictional mountain-refuge handover.
What to try & what it leaves out
Try. Remove a review, alter an evidence file or repeat a signer. Compare a complete record with a justified handover decision.
Limits. Event rules and file hashes check declared records and bytes. They do not authenticate people, prove independence or establish engineering acceptance.
Earlier example, repaired and reviewed 2 October 2026. This review checks the stated model and interactions; it does not establish usefulness in practice.
GitHub source (may require access)
Interactive example
Scenario and control lattice
Group fictional go-live scenarios by shared control and evidence tags using Formal Concept Analysis.
What to try & what it leaves out
Try. Inspect the four scenarios sharing a training tag, then reveal evidence-only and empty rows.
Limits. Tag presence does not prove an effective control. Shared attributes need not mean identical rows; an empty row may mean missing data.
Earlier example, repaired and reviewed 3 October 2026. This review checks the stated model and interactions; it does not establish usefulness in practice.
GitHub source (may require access)
Interactive example
Find your way into project controls
Explore the discipline’s relationships and a separate eight-step learning route for a project manager.
What to try & what it leaves out
Try. Search or filter, inspect a concept and its real links, then follow the learning route.
Limits. Curated concepts and predicates. Teaching order is a suggested reading path, not an additional set of graph relationships.
Earlier example, repaired and reviewed 2 October 2026. This review checks the stated model and interactions; it does not establish usefulness in practice.
GitHub source (may require access)
Interactive example
Project graph views
Generate parameterised views from a typed project graph and inspect why each item was included.
What to try & what it leaves out
Try. Run two views, change a parameter and inspect the selected items, relationships and inclusion reasons before exporting.
Limits. Queries over supplied records, not automatic project reasoning. Export formats require mapping and validation in the receiving tool.
Earlier example, repaired and reviewed 2 October 2026. This review checks the stated model and interactions; it does not establish usefulness in practice.
GitHub source (may require access)
Earlier workflow · 2019
Turn project records into a question worth testing
Data preparation branches into model comparison, visual exploration and an intended prediction route.
What this example leaves open
The target is IEG Bank Performance, not general project success. The workbook is absent; this is not a reproduced early forecast or working deployment.
Historical article ·
Applying appropriate machine-learning approach.
An earlier route from a project use case to a method, model, environment and dataset.
Interactive example
Read a crossing project through its notices
Connect project concepts, organisations, approximate locations and dated procurement notices.
What to try & what it leaves out
Try. Select a notice and compare its recorded facts, source check, map location and network links.
Limits. A bounded historical snapshot with some unverified rows; office points are not project geometry or evidence of a contract award.
Earlier example, repaired and reviewed 3 October 2026. This review checks the stated model and interactions; it does not establish usefulness in practice.
GitHub source (may require access)
Interactive example
Reading claims from project text
Inspect the step from draft annotations and selected statements to themes, stance and intention.
What to try & what it leaves out
Try. Search for an annotation or a statement, then compare its original label with what the text actually supports.
Limits. Unverified draft annotations and historical supplied scores; neither example establishes adoption trends or representative opinion.
Earlier example, repaired and reviewed 3 October 2026. This review checks the stated model and interactions; it does not establish usefulness in practice.
GitHub source (may require access)
Interactive example
Read project health and benefit assumptions
Inspect six project gap profiles, or explore a separate fictional programme’s benefit, consent, evidence and interface assumptions.
What to try & what it leaves out
Try. Explain a health tile from its three gap values, then switch to the independent benefits rehearsal and inspect a saved scenario.
Limits. Two distinct teaching models. The gaps and heuristic scores are supplied judgements; a scenario snapshot is not evidence of approval, verified assurance or realised benefits.
Earlier example, repaired and reviewed 3 October 2026. This review checks the stated model and interactions; it does not establish usefulness in practice.
GitHub source (may require access)