The data | Initial cohorts | Fit a Process-FINN | Replace growth with a neural network (Hybrid-FINN) | Training convergence | What the model learned: summary() | Assess the fit: in-sample and on held-out sites | Interpreting the growth process with ALE | Environmental niches | Which drivers matter: feature_importance() | Response to size and light: did the network recover the mechanistic form?
Package: FINN 0.2.0
Yannek Käber
FINN: Forest Informed Neural Networks
A hybrid dynamic forest (gap) model (FINN) that can be configured as a fully mechanistic, process-based model, like classic forest gap models, or with its demographic processes (growth, mortality, regeneration) replaced by deep neural networks (DNNs), or any combination of the two. Provides functions to define a model and its mechanistic or empirical components, calibrate it to forest inventory data, and interpret the calibrated processes. FINN is implemented with the 'torch' package, which supplies GPU support and the automatic differentiation used to calibrate the model by stochastic gradient descent; no knowledge of 'torch' is required. The hybrid modeling approach is described in Pichler and Käber (2026) <doi:10.1111/2041-210x.70347>.
Authors:
FINN_0.2.0.tar.gz
FINN_0.2.0.zip(r-4.7-x86_64)FINN_0.2.0.zip(r-4.7-arm64)FINN_0.2.0.zip(r-4.6-x86_64)FINN_0.2.0.zip(r-4.6-arm64)FINN_0.2.0.zip(r-4.5-x86_64)
FINN_0.2.0.tgz(r-4.6-x86_64)FINN_0.2.0.tgz(r-4.6-arm64)FINN_0.2.0.tgz(r-4.5-x86_64)FINN_0.2.0.tgz(r-4.5-arm64)
FINN_0.2.0.tar.gz(r-4.7-arm64)FINN_0.2.0.tar.gz(r-4.7-x86_64)FINN_0.2.0.tar.gz(r-4.6-arm64)FINN_0.2.0.tar.gz(r-4.6-x86_64)
FINN_0.2.0.tgz(r-4.6-emscripten)
manual.pdf |manual.html✨
DESCRIPTION |NEWS
card.svg |card.png
FINN/json (API)
| # Install 'FINN' in R: |
| install.packages('FINN', repos = c('https://finnverse.r-universe.dev', 'https://cloud.r-project.org')) |
Bug tracker:https://github.com/finnverse/finn/issues
Pkgdown/docs site:https://finnverse.github.io
Datasets:
- fia_env_dt - Oregon FIA calibration data: site climate
- fia_env_test - Oregon FIA calibration data: site climate
- fia_init_test - Oregon FIA calibration data: initial tree list
- fia_init_trees - Oregon FIA calibration data: initial tree list
- fia_obs_dt - Oregon FIA calibration data: training sites
- fia_obs_test - Oregon FIA calibration data: training sites
- fia_species_dt - Oregon FIA calibration data: species coding
Last updated from:609a30c72e. Checks:14 OK, 1 ERROR. Indexed: yes.
| Target | Result | Time | Files | Syslog |
|---|---|---|---|---|
| linux-devel-arm64 | OK | 163 | ||
| linux-devel-x86_64 | OK | 179 | ||
| source / vignettes | ERROR | 279 | ||
| linux-release-arm64 | OK | 148 | ||
| linux-release-x86_64 | OK | 170 | ||
| macos-release-arm64 | OK | 196 | ||
| macos-release-x86_64 | OK | 359 | ||
| macos-oldrel-arm64 | OK | 215 | ||
| macos-oldrel-x86_64 | OK | 258 | ||
| windows-devel-arm64 | OK | 474 | ||
| windows-devel-x86_64 | OK | 189 | ||
| windows-release-arm64 | OK | 414 | ||
| windows-release-x86_64 | OK | 156 | ||
| windows-oldrel-x86_64 | OK | 141 | ||
| wasm-release | OK | 141 |
Exports:aggregate_results_oldALEapply_env_scalingarray2obsDFaverageConditionalEffectsBA_standBA_stembinomial_from_bernoullibinomial_from_gammaclimateDF2arrayCohortMatcompetitioncompute_env_scalingconditionalEffectscreateHybridcreateProcessdbh2bafeature_importancefinnFINN.seedfitgroupby_meangrowthheightindex_speciesmakeInitCohortsmakeObsDatamortalitynp_runifobsDF2arrayspred2DFregenerationregeneration_saturationresolveSiteIDsrweibull_cohortssample_poisson_relaxedsimulateForest
Dependencies:abindbitbit64callrclicorocpp11data.tabledescfarverggplot2gluegtableisobandjsonlitelabelinglifecyclemagrittrotelprocessxpsR6RColorBrewerRcpprlangS7safetensorsscalestorchvctrsviridisLitewithr
Vignettes and/or manual for this package failed to build, some documents may be unavailable. Please inspect thebuild logs for more information.
Fitting FINN to forest inventory data (Oregon, US FIA)
Last update: 2026-09-22
Started: 2026-07-14
Mortality: a binomial response and a neural-network process
Mortality is a proportion, not a rate | Why binomial and not mse | Two models: mechanistic mortality, and mortality as a network | Scoring mortality: why not Spearman | What the network learned: ALE
Last update: 2026-09-22
Started: 2026-07-14
Introduction to FINN
How FINN works | Installation | Run the model | Species parameters | Environment and disturbances | Simulate
Last update: 2026-08-11
Started: 2026-07-14
Parameterising FINN from Ellenberg indicator values
From Ellenberg to FINN parameters | Simulate succession | The successional pattern | Ecological plausibility of the successional pattern | Adding a climate gradient | References
Last update: 2026-07-28
Started: 2026-07-14
Preparing your data for FINN
What FINN needs | Start from a raw tree list | Acquiring raw data (example: US FIA) | Define status flags | Build obs_dt with makeObsData() | Environmental data | Resolve IDs across tables | Initial cohorts | Checklist before fitting
Last update: 2026-07-28
Started: 2026-07-14
