Last update: 2026-07-19
Version: 0.6.5 – Matterhorn2
About
The proximetricsR package provides tools for developing, validating, and deploying quantitative chemometric models for near-infrared (NIR) spectroscopy, with dedicated support for BUCHI NIR sensors and workflows.
The package implements partial least squares (PLS) regression and related methods together with spectral preprocessing, model validation, visualisation, and native support for BUCHI calibration and application formats.
Installation
Install the development version from GitHub:
# install.packages("remotes")
remotes::install_github("l-ramirez-lopez/proximetricsr")Requires R ≥ 4.2.0 and compilation tools (C++ via Rcpp and RcppArmadillo).
A couple of examples
library(proximetricsR)
data("NIRcannabis")
# the list of formulas for the models to be built
app_formulas <- list(THC ~ spc, THCA ~ spc, CBD ~ spc, CBDA ~ spc)
# the list of pre-processing recipes to be tested
precipes <- list(
recipe_1 = preprocess_recipe(
prep_resample(grid = c(1001, 1700, 2)),
prep_snv(),
prep_derivative(m = 1, w = 9, p = 7, algorithm = "nwp"),
device = "proximate"
),
recipe_2 = preprocess_recipe(
prep_resample(grid = c(1001, 1700, 2)),
prep_snv(),
prep_derivative(m = 2, w = 11, p = 9, algorithm = "nwp"),
device = "proximate"
)
)
optimized_app <- calibrate_models(
formulas = app_formulas,
data = NIRcannabis,
preprocess_recipes = precipes,
methods = list(fit_plsr(15, type = "nwp")),
return_inputs = TRUE,
save_all = FALSE
)
optimized_appGrid search results:
formula recipe min property max property ncomp rsq rmse
1 THC ~ spc 1 0.0136 5.44 2 0.532 0.679
2 * THC ~ spc 2 0.0136 5.44 3 0.690 0.578
3 THCA ~ spc 1 0.0200 12.13 7 0.780 1.539
4 * THCA ~ spc 2 0.0200 12.13 6 0.820 1.407
5 CBD ~ spc 1 0.0262 5.89 1 0.148 0.675
6 * CBD ~ spc 2 0.0262 5.89 1 0.175 0.668
7 CBDA ~ spc 1 0.0100 24.64 4 0.684 3.454
8 * CBDA ~ spc 2 0.0100 24.64 2 0.682 3.431
largest_residual rsq_sd rmse_sd largest_residual_sd outliers method
1 1.87 0.1112 0.171 0.978 0 PLS (nwp)
2 1.67 0.1025 0.191 1.047 0 PLS (nwp)
3 4.27 0.1033 0.366 1.323 0 PLS (nwp)
4 3.63 0.0726 0.265 0.944 0 PLS (nwp)
5 2.17 0.0849 0.301 1.599 0 PLS (nwp)
6 2.18 0.0734 0.300 1.573 0 PLS (nwp)
7 11.37 0.2049 1.312 6.147 0 PLS (nwp)
8 11.66 0.2310 1.477 6.749 0 PLS (nwp)
*best model
---
Suggested models:
Model: THC ~ spc
Spectral preprocessing recipe (device: "proximate"):
- Step 1: prep_resample
min_wav: 1001; max_wav: 1700; resolution: 2
- Step 2: prep_snv
- Step 3: prep_derivative
m: 2; w: 11; p: 9; algorithm: 'nwp'
Method: PLS (nwp)
Model: THCA ~ spc
Spectral preprocessing recipe (device: "proximate"):
- Step 1: prep_resample
min_wav: 1001; max_wav: 1700; resolution: 2
- Step 2: prep_snv
- Step 3: prep_derivative
m: 2; w: 11; p: 9; algorithm: 'nwp'
Method: PLS (nwp)
Model: CBD ~ spc
Spectral preprocessing recipe (device: "proximate"):
- Step 1: prep_resample
min_wav: 1001; max_wav: 1700; resolution: 2
- Step 2: prep_snv
- Step 3: prep_derivative
m: 2; w: 11; p: 9; algorithm: 'nwp'
Method: PLS (nwp)
Model: CBDA ~ spc
Spectral preprocessing recipe (device: "proximate"):
- Step 1: prep_resample
min_wav: 1001; max_wav: 1700; resolution: 2
- Step 2: prep_snv
- Step 3: prep_derivative
m: 2; w: 11; p: 9; algorithm: 'nwp'
Method: PLS (nwp) Core functionality
Spectral preprocessing
A collection of preprocessing methods is available through functions with the prep_* prefix. The preprocess_recipe() framework enables the construction of reproducible preprocessing pipelines that can be applied consistently during calibration and prediction.
Supported operations include:
- Spectral resampling
- Savitzky-Golay smoothing and derivatives
- Standard normal variate (SNV)
- Detrending
- Gap-segment derivatives
- Transformations and scaling
Model calibration
The package supports multiple regression approaches, including:
- Partial least squares (PLS)
- Modified PLS (MPLS)
- NIRWise PLUS-compatible workflows
- XLS variants
The calibrate_models() function enables automated calibration, parameter optimisation, and model comparison using a range of cross-validation strategies.
Model validation and visualisation
Built-in tools support:
Model performance assessment
Outlier detection
Uncertainty estimation
Model interpretation
Publication-ready diagnostic plots can be generated using plot.spectral_model().
BUCHI ProxiMate & ProxiScout integration
proximetricsR implements algorithms that are numerically consistent with BUCHI NIRWise PLUS software, enabling reproducible workflows between R and production BUCHI devices.
Key capabilities include:
Native file I/O: Read and write ProxiMate and ProxiScout formats (
.cal,.prj,.nax,.rtf,.nad,.tsv)Application bundles: Create deployable
.naxapplications containing calibration models, preprocessing metadata, and device-specific parametersSensor-aware workflows: Support preprocessing and calibration workflows tailored to BUCHI sensor characteristics
Cross-software reproducibility: Develop and validate models in R and deploy them directly to BUCHI instruments without re-parameterisation
Documentation
For a complete overview of available vignettes:
browseVignettes("proximetricsR")Additional documentation can be accessed through:
help(package = "proximetricsR")References
Wold, S. (1975). Pattern recognition by means of disjoint principal components models. Pattern Recognition, 8(3), 127–139. doi:10.1016/B978-0-12-103950-9.50017-4
Shenk, J. S., & Westerhaus, M. O. (1991). The application of near infrared reflectance spectroscopy (NIRS) to compositional analysis of agricultural products. Crop Science, 31(2), 409–413. doi:10.2135/cropsci1991.0011183X003100020049x
Westerhaus, M. O. (2014). Eastern Analytical Symposium Award for outstanding innovations in near infrared spectroscopy. NIR News, 25(7), 7–10. doi:10.1255/nirn.1492
Support
For bug reports, feature requests, and discussions, please create an issue on GitHub:
https://github.com/buchi-labortechnik-ag/proximetricsR/issues
For questions, contact: