HPI: A Flagship Example of Explainable Index Creation
The Human Performance Index demonstrates how DataUniversa can transform separate, structurally different datasets into one comparable measure without losing the identity, context, or provenance of the observations beneath it.
Three sources. One explainable index.
HPI draws from three different forms of human-performance evidence. Each source follows a different structure, but DataUniversa allows them to contribute to a shared analytical framework.
GFF Standard
The same four-exercise universal test, with completion time as the principal performance measure.
GFF Pro
A separate standardized test designed around a higher level of difficulty and performance.
Local Records
Different exercises are interpreted within appropriate comparison pools and converted into comparable evidence.
A score of 100 identifies the leader within a defined age and gender comparison category. The score shows category leadership; the underlying pathway explains how it was achieved.
From raw observations to an explainable final score
HPI is not created by forcing every participant into an identical test pathway. Instead, DataUniversa progressively adds structure. Raw performances become identifiable records, those records contribute to source-specific indexes, and the source indexes are recombined into a higher-order HPI before contextual adjustment.
One participant may have an HPI based entirely on GFF Standard, another entirely on GFF Pro, another on Local Records, and another on a combination of all three. The final measure remains comparable because each source is transformed through an explicit calculation chain rather than hidden inside a black box.
Raw observations
Tests, exercises, results, dates, and participant evidence.
Structured records
Identity, context, location, demographics, and methodology.
Source indexes
GFFI and LRI convert source-specific evidence into comparable measures.
Unadjusted HPI
The contributing source indexes are recombined.
Contextual adjustment
Comparison occurs within the relevant age and gender pool.
Adjusted HPI
The category leader receives an adjusted score of 100.
A winners table that explains more than performance
The flagship HPI spreadsheet selects the leader in every defined age and gender category. Because each listed participant is the top performer in the relevant comparison group, every winner receives an Adjusted HPI of 100.
The meaningful comparison is therefore not the final adjusted score itself. It is the evidentiary pathway behind the score. Different winners may reach the same category-leading result through completely different combinations of GFF Standard, GFF Pro, and Local Records evidence.
Pool size remains visible
A winner among four participants and a winner among 253 participants can both legitimately receive an Adjusted HPI of 100. The HPI communicates leadership within the defined category, while the Participant Pool Count separately communicates the breadth of the competition.
These concepts remain separate so contextual breadth does not distort the meaning of the index itself.
Each layer adds knowledgewithout erasing what came before
DataUniversa does not merely aggregate data. It creates a hierarchy in which every successive layer adds analytical value while preserving the evidence beneath it. That makes HPI both explainable and reversible.
Raw observations
A participant completes GFF Standard, GFF Pro, or an individual Local Record exercise.
Structured performance records
The result is connected to the test, participant, age, gender, location, date, and relevant context.
Source-specific indexes
Standardized tests contribute to GFFI, while heterogeneous Local Records contribute to LRI.
Recombinant index
GFFI and LRI are combined into an Unadjusted Human Performance Index.
Contextual adjustment
The unadjusted score is compared within the appropriate age and gender category.
Explainability and reversibility
Every final score can be traced back to the exact observations that produced it.
Recombination should never destroy provenance
The final HPI is not a black-box number. It can be decomposed into the exact tests, exercises, source scores, indexes, demographic category, and participant pool that contributed to it. Explainability is therefore not added after the calculation; it is built into the architecture from the beginning.
The complete calculation chain remains visible and auditable, allowing a higher-order conclusion to remain permanently connected to the observations from which it was derived.
A model for future index creation
HPI matters because the architecture is transferable. The same approach can be applied anywhere heterogeneous observations must be transformed into a common but explainable measure. Different evidence types can retain their identity, contribute through source-specific indexes, and become part of a new higher-order measure without losing provenance.
Health
Education
Employee Performance
Agriculture
Rehabilitation
Economic Development
Product Quality
AI Systems
HPI is not only a fitness index. It is a working demonstration of the DataUniversa architecture.
Through structured observations, source-specific indexes, explainable recombination, contextual adjustment, and preserved provenance, DataUniversa can create new measures of value that did not exist in any original dataset while keeping every conclusion traceable to its source.