Core responsibility
PreciseDist is the core relatedness package. Its job is to make the numerical representation of object relatedness explicit, broad, typed, and inspectable.
The package boundary should include functions that create or change relatedness matrices:
-
precise_dist()builds many native distance and similarity matrices. -
precise_transform()performs explicit type conversion and matrix hygiene. -
precise_correlations()measures agreement among relatedness matrices. -
precise_stability()resamples the candidate matrix set to test metric-set sensitivity. -
precise_fusion()reduces many typed matrices to one fused matrix. -
precise_graph()projects typed matrices into graph-derived similarity matrices.
This is the intellectual center of the project. Visualization matters because seeing structure matters, but the first-class object is still the matrix: what it measures, what type it is, where it came from, and whether alternative metrics agree.
Boundary rule
PreciseDist should not become a general clustering framework or a general plotting package. It should provide the relatedness objects that those tools can consume.
The core package should stay focused on:
- distance and similarity breadth;
- explicit type metadata;
- conservative transformations;
- agreement diagnostics;
- metric-set stability diagnostics;
- matrix fusion;
- graph projection as matrix output;
- resource-aware execution.