repositoryrepositoryPortable nirs4all-core baseline using only cross-language supported operators.
unvalidatedThis recipe is the repository-side fixture for the V1 provider/repository/core roundtrip gate. It deliberately uses only operators in the current nirs4all-core portable subset:
KennardStoneSplitterStandardNormalVariateSavitzkyGolayPLSRegressionThe recipe is not a benchmark winner. Its job is to prove that the repository can serve a static pipeline descriptor that thin provider clients can resolve and core language bindings can load without importing the Python provider layer.
{
"name": "Core portable SNV + Savitzky-Golay + PLS",
"description": "nirs4all-core portable baseline for provider/repository/core roundtrip gates.",
"random_state": 42,
"pipeline": [
{
"class": "nirs4all.operators.splitters.KennardStoneSplitter",
"params": {
"test_size": 0.25
}
},
{
"class": "nirs4all.operators.transforms.StandardNormalVariate"
},
{
"class": "nirs4all.operators.transforms.SavitzkyGolay",
"params": {
"window_length": 11,
"polyorder": 3,
"deriv": 0,
"mode": "interp",
"cval": 0.0
}
},
{
"model": {
"class": "sklearn.cross_decomposition.PLSRegression"
},
"param": "n_components",
"_range_": [
2,
11,
2
]
}
]
}
# Python
import nirs4all_repository as n4r
pipe = n4r.get("core_portable_snv_savgol_pls")
config = pipe.to_nirs4all() # ready for nirs4all.run() / predict()# any language: read the index, fetch + verify
curl https://repository.nirs4all.org/data/index.json
curl https://repository.nirs4all.org/data/pipelines/core_portable_snv_savgol_pls/pipeline.json