{ "_id": "6ad8adb0c230b829876fcf36", "Package": "soilcnn", "Title": "Convolutional Neural Networks for Digital Soil Mapping", "Version": "0.1.0", "Authors@R": "c(\nperson(c(\"Cássio\", \"Marques\"), \"Moquedace\", , \"cassiomoquedace@gmail.com\",\nrole = c(\"aut\", \"cre\", \"cph\")),\nperson(c(\"Clara\", \"Glória\", \"Oliveira\"), \"Baldi\", role = c(\"aut\", \"cph\")))", "Description": "A framework, built on 'torch', for mapping a soil property\nfrom point observations and a stack of aligned rasters with a\nconvolutional network that reads the neighbourhood of each\npoint. It extracts the patches; resamples by spatial blocks or\nby k-fold nearest neighbour distance matching, after\nLinnenbrink et al. (2024) ; tunes\nand selects under a noise floor and the one-standard-error\nrule; refits the selected configuration under several seeds;\nand predicts median, mean and interval maps over rasters of any\nextent. The map's centre is the ensemble median over seeds; the\nmean corrects its back-transformation by smearing, after Duan\n(1983) . The intervals are\nconformal, calibrated three ways and each checked on the test\nset: on the cross-validated residuals; on a calibration set\nheld out of every fold, as split conformal, after Lei et al.\n(2018) ; or with every\nfold's model at each pixel, as CV+, after Barber et al. (2021)\n -- with points, or whole groups of\nthem, weighing alike, after Dunn et al. (2023)\n. The area of applicability\nof Meyer and Pebesma (2021) is\nmapped beside them. What a model learned is read by\npermutation, by SHAP values from expected gradients, after\nErion et al. (2021) , by SAGE,\nafter Covert et al. (2020) , by\naccumulated local effects, after Apley and Zhu (2020)\n, and by refits without each variable.\nRandom forest and multilayer perceptron baselines run under the\nsame folds.", "License": "MIT + file LICENSE", "URL": "https://github.com/moquedace/soilcnn,\nhttps://moquedace.github.io/soilcnn/", "BugReports": "https://github.com/moquedace/soilcnn/issues", "Config/testthat/edition": "3", "VignetteBuilder": "knitr", "Encoding": "UTF-8", "Language": "en-GB", "Config/roxygen2/version": "8.1.0", "NeedsCompilation": "no", "Packaged": { "Date": "2026-10-21 12:14:30 UTC", "User": "root" }, "Author": "Cássio Marques Moquedace [aut, cre, cph], Clara Glória Oliveira\nBaldi [aut, cph]", "Maintainer": "Cássio Marques Moquedace ", "Repository": "https://cran.r-universe.dev", "Date/Publication": "2026-10-21 10:19:38 UTC", "RemoteUrl": "https://github.com/cran/soilcnn", "RemoteRef": "HEAD", "RemoteSha": "2379c381d2cc1dea08270023c35fab0d5565455e", "_user": "cran", "_type": "src", "_file": "soilcnn_0.1.0.tar.gz", "_fileid": "https://r2.ropensci.org/522bdd57ea7abbeb16458c93200dfac5b21ec21511db330636c0254fdcf1ca8d", "_filesize": 3567167, "_sha256": "522bdd57ea7abbeb16458c93200dfac5b21ec21511db330636c0254fdcf1ca8d", "_expires": "2027-01-29T12:18:50.000Z", "_created": "2026-10-21T12:14:30.000Z", "_published": "2026-10-21T12:18:56.410Z", "_upstream": "https://github.com/cran/soilcnn", "_status": "success", "_maintainer": { "name": "Cássio Marques Moquedace", "email": "cassiomoquedace@gmail.com", "login": "moquedace", "twitter": "@CMoquedace", "description": "Agroecologist • Forestry Engineer • Ph.D. in Soil Science • Postdoctoral Researcher at ESALQ/USP 🌱 🍂🌳📏🌎", "uuid": 67038094 }, "_jobs": [ { "job": 119215390422, "time": 198, "config": "linux-devel-x86_64", "r": "4.7.0", "check": "OK", "artifact": "12344533175" }, { "job": 119215390400, "time": 208, "config": "linux-release-x86_64", "r": "4.6.1", "check": "OK", "artifact": "12345081392" }, { "job": 119213084654, "time": 310, "config": "source", "r": "4.6.1", "check": "OK", "artifact": "12344841298" }, { "job": 119215390262, "time": 144, "config": "wasm-release", "r": "4.6.0", "check": "OK", "artifact": "12344826356" }, { "job": 119215390367, "time": 146, "config": "windows-devel-x86_64", "r": "4.7.0", "check": "OK", "artifact": "12344488147" }, { "job": 119215390339, "time": 146, "config": "windows-release-x86_64", "r": "4.6.1", "check": "OK", "artifact": "12344846350" } ], "_buildurl": "https://github.com/r-universe/cran/actions/runs/39597366527", "_commit": { "id": "2379c381d2cc1dea08270023c35fab0d5565455e", "author": "Cássio Marques Moquedace ", "committer": "cran-robot ", "message": "version 0.1.0\n", "time": 1792577978 }, "_distro": "resolute", "_host": "GitHub-Actions", "_registered": true, "_dependencies": [ { "package": "R", "version": ">= 4.5.0", "role": "Depends" }, { "package": "callr", "role": "Imports" }, { "package": "coro", "role": "Imports" }, { "package": "dplyr", "role": "Imports" }, { "package": "graphics", "role": "Imports" }, { "package": "grDevices", "role": "Imports" }, { "package": "janitor", "role": "Imports" }, { "package": "matrixStats", "role": "Imports" }, { "package": "parallel", "role": "Imports" }, { "package": "ps", "role": "Imports" }, { "package": "purrr", "role": "Imports" }, { "package": "readr", "role": "Imports" }, { "package": "stats", "role": "Imports" }, { "package": "terra", "role": "Imports" }, { "package": "tibble", "role": "Imports" }, { "package": "tools", "role": "Imports" }, { "package": "torch", "role": "Imports" }, { "package": "utils", "role": "Imports" }, { "package": "caret", "role": "Suggests" }, { "package": "CAST", "role": "Suggests" }, { "package": "FNN", "role": "Suggests" }, { "package": "knitr", "role": "Suggests" }, { "package": "pkgload", "role": "Suggests" }, { "package": "randomForest", "role": "Suggests" }, { "package": "ranger", "role": "Suggests" }, { "package": "rmarkdown", "role": "Suggests" }, { "package": "sf", "role": "Suggests" }, { "package": "testthat", "version": ">= 3.0.0", "role": "Suggests" } ], "_owner": "cran", "_selfowned": false, "_usedby": 0, "_updates": [ { "week": "2026-43", "n": 1 } ], "_tags": [ { "name": "0.1.0", "date": "2026-10-21" } ], "_stars": 0, "_contributors": [ { "user": "moquedace", "count": 1, "uuid": 67038094 } ], "_userbio": { "uuid": 6899542, "type": "organization", "name": "cran", "followers": 634, "description": "Unofficial read-only mirror of all CRAN R packages" }, "_downloads": { "count": 0, "source": "https://cranlogs.r-pkg.org/downloads/total/last-month/soilcnn" }, "_devurl": "https://github.com/moquedace/soilcnn", "_pkgdown": "https://moquedace.github.io/soilcnn/", "_searchresults": 0, "_rbuild": "4.6.1", "_assets": [ "extra/citation.cff", "extra/citation.html", "extra/citation.json", "extra/citation.txt", "extra/contents.json", "extra/manual.pdf", "extra/NEWS.html", "extra/NEWS.txt", "extra/soilcnn.html", "LICENSE" ], "_cranurl": false, "_exports": [ "ale_effect", "aoa_di", "aoa_reference", "aoa_threshold", "block_bootstrap", "block_bootstrap_by_size", "calc_metrics", "caret_available", "caret_spec", "ccc", "check_fold_plan", "compare_importance", "conformal_calibrate", "conformal_cv", "conformal_interval", "conformal_scale_fit", "conformal_scaled_calibrate", "conformal_scaled_interval", "context_importance", "cv_plus_calibrate", "cv_plus_interval", "cv_residuals", "cv_residuals_for_config", "di_reference", "dissimilarity_index", "dsm_final", "dsm_importance", "dsm_load", "dsm_predict", "dsm_prepare", "dsm_report_final", "dsm_train", "equal_area_blocks", "example_landscape", "example_run", "fold_leakage_report", "freeze_selection", "get_model", "holdout", "holdout_cv", "importance_groups", "importance_map", "importance_points", "importance_weights", "inside_aoa", "knndm_cv", "knndm_folds", "latest_run_dir", "list_models", "make_manual_tune_grid", "make_tune_grid", "model_spec", "occlusion_report", "one_se", "paired_family_test", "permutation_importance", "picp_report", "prediction_sample", "print_aoa", "print_noise_floor", "print_one_se", "random_cv", "random_folds", "refit_importance", "region_cv", "region_folds", "register_model", "resolve_resampling", "sage_importance", "score_test_grid", "seed_noise_floor", "selected_config_id", "set_torch_threads", "setup_torch_device", "shap_importance", "smear", "smear_map", "smearing_check", "smearing_factor", "smearing_from_run", "spatial_correlogram", "spatial_cv", "spatial_folds", "summarise_resamples", "target_transform_spec" ], "_help": [ { "page": "ale_effect", "title": "ALE: how the prediction changes along each variable's range.", "topics": [ "ale_effect" ] }, { "page": "aoa_di", "title": "The DI of new rows of RAW predictor values, QC'd and scaled as the reference was.", "topics": [ "aoa_di" ] }, { "page": "aoa_reference", "title": "The dissimilarity reference of a fitted model.", "topics": [ "aoa_reference" ] }, { "page": "aoa_threshold", "title": "The DI threshold that separates the area of applicability.", "topics": [ "aoa_threshold" ] }, { "page": "block_bootstrap", "title": "A metric, or two models' difference in it, with an interval from whole blocks.", "topics": [ "block_bootstrap" ] }, { "page": "block_bootstrap_by_size", "title": "The interval as the blocks grow.", "topics": [ "block_bootstrap_by_size" ] }, { "page": "calc_metrics", "title": "Compute all regression metrics for one obs/pred pair.", "topics": [ "calc_metrics" ] }, { "page": "caret_available", "title": "Every caret method that could be borrowed here.", "topics": [ "caret_available" ] }, { "page": "caret_spec", "title": "Turn a caret method into a model_spec.", "topics": [ "caret_spec" ] }, { "page": "ccc", "title": "Lin's Concordance Correlation Coefficient.", "topics": [ "ccc" ] }, { "page": "check_fold_plan", "title": "Check a plan's folds are well formed, and describe them.", "topics": [ "check_fold_plan" ] }, { "page": "compare_importance", "title": "Several importances, side by side.", "topics": [ "compare_importance" ] }, { "page": "conformal_calibrate", "title": "Calibrate a conformal interval from held-out residuals.", "topics": [ "conformal_calibrate" ] }, { "page": "conformal_cv", "title": "Calibrate on the folds of a resampling run, then report coverage.", "topics": [ "conformal_cv" ] }, { "page": "conformal_interval", "title": "Turn predictions into intervals.", "topics": [ "conformal_interval" ] }, { "page": "conformal_scale_fit", "title": "Fit the scale of an interval: |residual| on covariates that say how wrong.", "topics": [ "conformal_scale_fit" ] }, { "page": "conformal_scaled_calibrate", "title": "Calibrate an interval whose width is a fitted scale.", "topics": [ "conformal_scaled_calibrate" ] }, { "page": "conformal_scaled_interval", "title": "Turn predictions into intervals with a fitted scale.", "topics": [ "conformal_scaled_interval" ] }, { "page": "context_importance", "title": "Context importance: how far from the point, and at which scale, the model reads.", "topics": [ "context_importance" ] }, { "page": "cv_plus_calibrate", "title": "Calibrate a CV+ interval from the folds of a cross-validation.", "topics": [ "cv_plus_calibrate" ] }, { "page": "cv_plus_interval", "title": "Turn the fold models' predictions into CV+ intervals.", "topics": [ "cv_plus_interval" ] }, { "page": "cv_residuals", "title": "Cross-validated residuals from a tuning run, for calibration.", "topics": [ "cv_residuals" ] }, { "page": "cv_residuals_for_config", "title": "Cross-validated residuals of a configuration, from any tuning run.", "topics": [ "cv_residuals_for_config" ] }, { "page": "di_reference", "title": "Summarise a training set for dissimilarity-index computation.", "topics": [ "di_reference" ] }, { "page": "dissimilarity_index", "title": "The dissimilarity index of new data.", "topics": [ "dissimilarity_index" ] }, { "page": "dsm_final", "title": "Refit the configuration a tuning run selects, under N seeds.", "topics": [ "dsm_final" ] }, { "page": "dsm_importance", "title": "Variable importance of a fitted model, by the method you choose.", "topics": [ "dsm_importance" ] }, { "page": "dsm_load", "title": "Load everything a run needs, and check that it fits together.", "topics": [ "dsm_load" ] }, { "page": "dsm_predict", "title": "Predict a map from a final model.", "topics": [ "dsm_predict" ] }, { "page": "dsm_prepare", "title": "Build a patch store from a point table and a folder of aligned rasters.", "topics": [ "dsm_prepare" ] }, { "page": "dsm_report_final", "title": "The declaration of a final model that already exists.", "topics": [ "dsm_report_final" ] }, { "page": "dsm_train", "title": "Fit and tune a model over a resampling plan.", "topics": [ "dsm_train" ] }, { "page": "equal_area_blocks", "title": "Square blocks of equal area, in kilometres.", "topics": [ "equal_area_blocks" ] }, { "page": "example_landscape", "title": "A small synthetic landscape, for the examples and the tests.", "topics": [ "example_landscape" ] }, { "page": "example_run", "title": "A small fitted run on the example landscape, made once a session.", "topics": [ "example_run" ] }, { "page": "fold_leakage_report", "title": "How much does each fold still leak?", "topics": [ "fold_leakage_report" ] }, { "page": "freeze_selection", "title": "Record which config was chosen, and when.", "topics": [ "freeze_selection" ] }, { "page": "get_model", "title": "Fetch a registered model.", "topics": [ "get_model" ] }, { "page": "holdout", "title": "A single train/validation/test split of a point table.", "topics": [ "holdout" ] }, { "page": "holdout_cv", "title": "A single train/validation/test split.", "topics": [ "holdout_cv" ] }, { "page": "importance_groups", "title": "Which channels the importance treats as one variable.", "topics": [ "importance_groups" ] }, { "page": "importance_map", "title": "Maps of SHAP values, from an importance computed at points of the map.", "topics": [ "importance_map" ] }, { "page": "importance_points", "title": "Points of the map to explain: a regular grid of the model's raster cells.", "topics": [ "importance_points" ] }, { "page": "importance_weights", "title": "One weight per channel, for the area of applicability, from an importance.", "topics": [ "importance_weights" ] }, { "page": "inside_aoa", "title": "Is each row inside the area of applicability?", "topics": [ "inside_aoa" ] }, { "page": "knndm_cv", "title": "Folds matched to where the map will be predicted (kNNDM).", "topics": [ "knndm_cv" ] }, { "page": "knndm_folds", "title": "Fold plan from k-fold Nearest Neighbour Distance Matching.", "topics": [ "knndm_folds" ] }, { "page": "latest_run_dir", "title": "The most recent run under 'base', by time, and only if it finished.", "topics": [ "latest_run_dir" ] }, { "page": "list_models", "title": "Every registered model, as a table.", "topics": [ "list_models" ] }, { "page": "make_manual_tune_grid", "title": "Build a manual grid from explicit lists of values per parameter.", "topics": [ "make_manual_tune_grid" ] }, { "page": "make_tune_grid", "title": "Generate a random hyperparameter grid (like caret's tuneLength).", "topics": [ "make_tune_grid" ] }, { "page": "model_spec", "title": "Describe a model the framework can fit.", "topics": [ "model_spec" ] }, { "page": "occlusion_report", "title": "What each part of the patch is worth to one trained unit of a tuning run.", "topics": [ "occlusion_report" ] }, { "page": "one_se", "title": "Pick the simplest config within one standard error of the best.", "topics": [ "one_se" ] }, { "page": "paired_family_test", "title": "Paired comparison between two model families.", "topics": [ "paired_family_test" ] }, { "page": "permutation_importance", "title": "Permutation importance: how much a fitted model relies on each variable.", "topics": [ "permutation_importance" ] }, { "page": "picp_report", "title": "Coverage, and whether it holds where it is needed.", "topics": [ "picp_report" ] }, { "page": "plot.block_bootstrap_by_size", "title": "Plot the interval as the blocks grow.", "topics": [ "plot.block_bootstrap_by_size" ] }, { "page": "plot.dsm_importance", "title": "Plot a variable importance.", "topics": [ "plot.dsm_importance" ] }, { "page": "plot.importance_comparison", "title": "Plot importances side by side.", "topics": [ "plot.importance_comparison" ] }, { "page": "plot.importance_map", "title": "Plot maps of SHAP values.", "topics": [ "plot.importance_map" ] }, { "page": "plot.spatial_correlogram", "title": "Plot a spatial correlogram.", "topics": [ "plot.spatial_correlogram" ] }, { "page": "prediction_sample", "title": "A regular sample of the prediction area, as kNNDM's 'predpoints'.", "topics": [ "prediction_sample" ] }, { "page": "print_aoa", "title": "Report what an AOA covers.", "topics": [ "print_aoa" ] }, { "page": "print_noise_floor", "title": "Print a noise-floor report in the terms it should be read in.", "topics": [ "print_noise_floor" ] }, { "page": "print_one_se", "title": "Say what one_se() did, including when it did nothing.", "topics": [ "print_one_se" ] }, { "page": "random_cv", "title": "Random k-fold. Ignores geography by construction.", "topics": [ "random_cv" ] }, { "page": "random_folds", "title": "Random k-fold over the train+validation pool.", "topics": [ "random_folds" ] }, { "page": "refit_importance", "title": "Refit importance: what the model loses when trained without a variable.", "topics": [ "refit_importance" ] }, { "page": "region_cv", "title": "Leave-region-out, on a grouping that already exists (biome, catchment, ...).", "topics": [ "region_cv" ] }, { "page": "region_folds", "title": "Fold by a grouping column: no group is split across folds.", "topics": [ "region_folds" ] }, { "page": "register_model", "title": "Add a model to the registry.", "topics": [ "register_model" ] }, { "page": "resolve_resampling", "title": "Turn a spec into a fold plan against real points.", "topics": [ "resolve_resampling" ] }, { "page": "sage_importance", "title": "SAGE: how much of the model's skill each variable carries, shared fairly.", "topics": [ "sage_importance" ] }, { "page": "score_test_grid", "title": "Score every trained unit of a tuning run on the held-out test set.", "topics": [ "score_test_grid" ] }, { "page": "seed_noise_floor", "title": "How much does a metric move when ONLY the seed changes?", "topics": [ "seed_noise_floor" ] }, { "page": "selected_config_id", "title": "The configuration a final run deployed.", "topics": [ "selected_config_id" ] }, { "page": "set_torch_threads", "title": "Set torch's thread pools.", "topics": [ "set_torch_threads" ] }, { "page": "setup_torch_device", "title": "Configure torch threads and select compute device.", "topics": [ "setup_torch_device" ] }, { "page": "shap_importance", "title": "SHAP importance: how each variable pushes each prediction, up or down.", "topics": [ "shap_importance" ] }, { "page": "smear", "title": "The conditional MEAN surface, from transform-space predictions.", "topics": [ "smear" ] }, { "page": "smear_map", "title": "The mean surface of a median map already made.", "topics": [ "smear_map" ] }, { "page": "smearing_check", "title": "The smearing factors on held-out points: which one unbiases the mean surface.", "topics": [ "smearing_check" ] }, { "page": "smearing_factor", "title": "The smearing factor, from held-out residuals in transform space.", "topics": [ "smearing_factor" ] }, { "page": "smearing_from_run", "title": "The smearing factor from a tuning run's out-of-fold predictions.", "topics": [ "smearing_from_run" ] }, { "page": "spatial_correlogram", "title": "Moran's I by distance class: how far apart two points stop resembling each other.", "topics": [ "spatial_correlogram" ] }, { "page": "spatial_cv", "title": "Spatially blocked k-fold: whole blocks of ground go to one fold.", "topics": [ "spatial_cv" ] }, { "page": "spatial_folds", "title": "Spatially blocked folds of a point table.", "topics": [ "spatial_folds" ] }, { "page": "summarise_resamples", "title": "Aggregate a unit-level comparison table into one row per config.", "topics": [ "summarise_resamples" ] }, { "page": "target_transform_spec", "title": "The forward and inverse functions for a named target transform.", "topics": [ "target_transform_spec" ] } ], "_rundeps": [ "bit", "bit64", "callr", "cli", "clipr", "coro", "cpp11", "crayon", "desc", "dplyr", "farver", "generics", "glue", "hms", "janitor", "jsonlite", "labeling", "lifecycle", "lubridate", "magrittr", "matrixStats", "otel", "pillar", "pkgconfig", "prettyunits", "processx", "progress", "ps", "purrr", "R6", "RColorBrewer", "Rcpp", "readr", "rlang", "safetensors", "scales", "snakecase", "stringi", "stringr", "terra", "tibble", "tidyr", "tidyselect", "timechange", "torch", "tzdb", "utf8", "vctrs", "viridisLite", "vroom", "withr" ], "_vignettes": [ { "source": "soilcnn.Rmd", "filename": "soilcnn.html", "title": "Getting started with soilcnn", "engine": "knitr::rmarkdown", "headings": [ "1. Prepare and load the patch store", "2. Understand the multiscale inputs", "3. Decide who trains and who scores", "4. Tune", "Tutorial example: choosing a supported level of complexity", "Application example: model selection", "5. The baselines that give the number a scale", "6. The test set, once", "Application example: validation designs on one test set", "7. Refit the chosen configuration", "Application example: coverage and interval width", "8. The map", "Where the map may be believed", "Application example: reading the final maps together", "9. What the model learned", "Application example: importance by theme", "A reference for your own run", "About the illustrations" ], "created": "2026-10-21 10:19:38", "modified": "2026-10-21 10:19:38", "commits": 1 } ], "_score": 2, "_indexed": false, "_nocasepkg": "soilcnn", "_universes": [ "cran", "moquedace" ], "_indexurl": "https://moquedace.r-universe.dev/soilcnn", "_binaries": [ { "r": "4.7.0", "os": "linux", "version": "0.1.0", "date": "2026-10-21T12:17:10.000Z", "distro": "resolute", "commit": "2379c381d2cc1dea08270023c35fab0d5565455e", "fileid": "https://r2.ropensci.org/3ab5d88014e2ede6786eda6dc1a8ab393eaba1c6262cde23d83f56e0f4f6d49f", "status": "success", "check": "OK", "buildurl": "https://github.com/r-universe/cran/actions/runs/39597366527" }, { "r": "4.6.1", "os": "linux", "version": "0.1.0", "date": "2026-10-21T12:17:25.000Z", "distro": "resolute", "commit": "2379c381d2cc1dea08270023c35fab0d5565455e", "fileid": "https://r2.ropensci.org/210c74fbbc5fb9baba86d1dfb71caa2e4dfc091ddf7c8a288b1cff47f1c9df31", "status": "success", "check": "OK", "buildurl": "https://github.com/r-universe/cran/actions/runs/39597366527" }, { "r": "4.6.0", "os": "wasm", "version": "0.1.0", "date": "2026-10-21T12:17:20.000Z", "commit": "2379c381d2cc1dea08270023c35fab0d5565455e", "fileid": "https://r2.ropensci.org/be1af4690d2c16c11cea5169b809ddf54de21f1488c869a8194617432b8d864a", "status": "success", "buildurl": "https://github.com/r-universe/cran/actions/runs/39597366527" }, { "r": "4.7.0", "os": "win", "version": "0.1.0", "date": "2026-10-21T12:16:05.000Z", "commit": "2379c381d2cc1dea08270023c35fab0d5565455e", "fileid": "https://r2.ropensci.org/5aedc61a293b4fc8239bfa367ffcc86a521892a0f75b59e572684fdf4775cbee", "status": "success", "check": "OK", "buildurl": "https://github.com/r-universe/cran/actions/runs/39597366527" }, { "r": "4.6.1", "os": "win", "version": "0.1.0", "date": "2026-10-21T12:16:23.000Z", "commit": "2379c381d2cc1dea08270023c35fab0d5565455e", "fileid": "https://r2.ropensci.org/015315fecb27bd58d472c552a2112d0c56ce8d902c6c1ed3662396e15d0d3794", "status": "success", "check": "OK", "buildurl": "https://github.com/r-universe/cran/actions/runs/39597366527" } ] }