filter_leverage — High Leverage Filter¶
Group: Sample / feature filters · Binding: n4m.sklearn.HighLeverageFilter · C ABI: n4m_filter_leverage_*
Description¶
Hat-matrix or PCA score-space leverage filter.
Parameters¶
Name |
Type |
Default |
|---|---|---|
|
`str |
int` |
|
|
|
|
`float |
None` |
|
|
|
|
|
|
Explanations¶
Bibliographic source¶
Standard spectroscopic operator — see the nirs4all preprocessing / augmentation handbook and the cited literature within the binding docstring.
Mathematical principle¶
Hat-matrix or PCA score-space leverage filter.
Implementation¶
C ABI n4m_filter_leverage_* in libn4m (create / apply / destroy lifecycle), wrapped by n4m.sklearn.HighLeverageFilter. The same numerical kernel backs every language binding.
Usage¶
from n4m.sklearn import HighLeverageFilter
op = HighLeverageFilter()
X_transformed = op.fit_transform(X)
Benchmarks¶
Adaptive wall-clock per cell measured against full_matrix.csv. Only backends that implement this method are listed; libraries without the method are omitted.
Verdict · ✓ ref / ≈ ref / ~ shape mark a reference-gate pass at strict / relaxed / qualitative tolerance · ✓ bind = pls4all binding agrees with the C++ baseline · ⇄ cross-check = documented by-design selector/RNG/model, noncanonical API/facade convention, or secondary oracle · ✗ divergent · ⚠ error · — not run. The fastest backend per column is marked 🏆.
Reference gate: strict — numeric equivalence (rmse_rel_tol ≤ 1e-12).
| Backend | Parity | 50×250 (ms) | 250×50 (ms) |
|---|---|---|---|
| C++ native · libn4m | |||
pls4all.cpp.blas | ✓ ref | — | 1.68 ms |
pls4all.cpp.blas+omp | ✓ ref | 415.7 ms | 2.39 ms |
pls4all.cpp.omp | ✓ ref | — | 1.54 ms |
pls4all.cpp.ref | ✓ ref | 387.8 ms | 1.57 ms |
| Python · pls4all | |||
pls4all.python | ✓ bind | 245.0 ms | 1.09 ms |
pls4all.sklearn | ✓ bind | 243.6 ms | 1.14 ms |
| Python · external | |||
nirs4all | source | 0.93 ms🏆 | 0.18 ms🏆 |
See also: methods index · interactive dashboard