mt op• Data kinds: matrix → table
• Call: import fullseye as fs; fs.ledger.spc_mt_unit_space(data) (to call the implementation directly, import spc; spc.spc_mt_unit_space(data); from the registry, opsspc.get("spc_mt_unit_space"))
Build an MT-method unit space from known-good observations.
`data is a 2-D array of shape (n, p) — n observations of p`
features, all of which must be normal product. The unit space is the
triple `(mean, std, inv_corr)`: per-feature mean and ddof=1 standard
deviation, and the inverse of the correlation matrix of the standardised
data. Feed it to :func:spc_mt_distance to score new observations.
Returns a dict with `mean, std, corr and inv_corr` arrays, the
unit space's own `md per row, md_sq_mean`, the exact expected value
`md_sq_mean_exact = (n-1)/n, the counts n and p`, how many features
had no spread (`flat_features`), and whether the correlation matrix was
singular enough to need a pseudo-inverse (`pseudo_inverse`).
The mean of `md**2 over the unit space equals (n-1)/n` exactly — the
quantitative form of "a unit space averages a distance of one". It reaches 1
only as `n grows; the deficit is exactly 1/n`.
Raises `ValueError`: a non-2-D / empty *data*, fewer than two
observations (no spread to standardise by), zero features, or non-finite
input.
• Sample-data catalog (download URLs / licences) — 2-D uses skimage.data (BSD/public domain) plus synthetic images; 3-D lists download URLs for real data sources (Stanford, PDS, …).
• Operator provenance and references — the sources of the research/methods this op family came from.
• The canonical algorithm (author, year) and its uses are named in the family usage guide above.
• poc_mt_hidden_fault — py -3.11 examples/poc_mt_hidden_fault.py
• poc_spc — py -3.11 examples/poc_spc.py
table as input)msa_anova_table · msa_gauge_rr · msa_bias_linearity · msa_attribute_agreement · gum_standard_uncertainty · gum_propagate · gum_expanded · gum_monte_carlo
mt)spc_mt_distance · spc_mt_sn_ratio
*Provenance: spc.py — SPC operator registry. This per-op note is generated by tools/opdocs.py md (do not hand-edit).*
© 2026 Kazufumi Furuse — Fullseye operator documentation. Licensed under Apache-2.0.