Merge pull request #78 from lanl/cv-testing

Merge updates in cv-testing branch to prepare for 0.6 release
lanl · Aug 4, 2022 · cc05ad2 · cc05ad2
2 parents 85ac198 + 6ca69db
commit cc05ad2
Show file tree

Hide file tree

Showing 17 changed files with 121 additions and 323 deletions.
diff --git a/DESCRIPTION b/DESCRIPTION
@@ -21,7 +21,7 @@ BugReports: https://github.com/lanl/NEONiso/issues
 URL: https://github.com/lanl/NEONiso
 Encoding: UTF-8
 LazyData: true
-RoxygenNote: 7.2.0
+RoxygenNote: 7.2.1
 Imports:
     dplyr,
     zoo,

diff --git a/NEWS.md b/NEWS.md
@@ -5,7 +5,7 @@ have some known issues (e.g., no correction is made for concentration dependence
 of the analyzers yet), and any data produced from this function should be considered
 provisional.
 * Added capability to plot data used in carbon calibration regression in order
-to help identify periods where calibration paramters seem to be okay, but
+to help identify periods where calibration parameters seem to be okay, but
 quality of calibrated data is degraded.
 * Added cross-validation error estimates to carbon calibration routines.
 * The calibrate_carbon_bymonth() function has been marked as deprecated, but will

diff --git a/R/calibrate_ambient_carbon_Bowling2003.R b/R/calibrate_ambient_carbon_Bowling2003.R
@@ -2,7 +2,7 @@
 #'
 #' @author Rich Fiorella \email{rfiorella@@lanl.gov}
 #'
-#' Function called by `calibrate_carbon_bymoth()` to apply
+#' Function called by `calibrate_carbon_bymonth()` to apply
 #' gain and offset parameters to the ambient datasets (000_0x0_09m and
 #' 000_0x0_30m). This function should generally not be used independently,
 #' but should be used in coordination with
@@ -155,8 +155,8 @@ calibrate_ambient_carbon_Bowling2003 <- function(amb_data_list,
   # extract 12CO2 and 13CO2 concentrations from the ambient data, 
   # set as NA, unless overwritten
 
-  mean12c <- max12c <- min12c <- cv5rmse12c <- loocv12c <- rep(NA, length(amb_delta$mean)) # placeholders for 12CO2 vecs
-  mean13c <- max13c <- min13c <- cv5rmse13c <- loocv13c <- rep(NA, length(amb_delta$mean)) # placeholders for 13CO2 vecs
+  mean12c <- max12c <- min12c <- cv5rmse12c <- cvloo12c <- rep(NA, length(amb_delta$mean)) # placeholders for 12CO2 vecs
+  mean13c <- max13c <- min13c <- cv5rmse13c <- cvloo13c <- rep(NA, length(amb_delta$mean)) # placeholders for 13CO2 vecs
 
   for (i in seq_len(length(var_inds_in_calperiod))) {
     # calculate calibrated 12CO2 concentrations
@@ -166,8 +166,8 @@ calibrate_ambient_carbon_Bowling2003 <- function(amb_data_list,
       amb_12CO2$min[var_inds_in_calperiod[[i]]] + caldf$offset12C[i]
     max12c[var_inds_in_calperiod[[i]]] <- caldf$gain12C[i] *
       amb_12CO2$max[var_inds_in_calperiod[[i]]] + caldf$offset12C[i]
-    cv5rmse12c[var_inds_in_calperiod[[i]]] <- caldf$cv5rmse_12C[i]
-    loocv12c[var_inds_in_calperiod[[i]]] <- caldf$loocv_12C[i]
+  #  cv5rmse12c[var_inds_in_calperiod[[i]]] <- caldf$cv5rmse_12C[i]
+  #  cvloo12c[var_inds_in_calperiod[[i]]] <- caldf$cvloo_12C[i]
 
     # calculate calibrated 13CO2 concentrations
     mean13c[var_inds_in_calperiod[[i]]] <- caldf$gain13C[i] *
@@ -176,8 +176,8 @@ calibrate_ambient_carbon_Bowling2003 <- function(amb_data_list,
       amb_13CO2$min[var_inds_in_calperiod[[i]]] + caldf$offset13C[i]
     max13c[var_inds_in_calperiod[[i]]] <- caldf$gain13C[i] *
       amb_13CO2$max[var_inds_in_calperiod[[i]]] + caldf$offset13C[i]
-    cv5rmse13c[var_inds_in_calperiod[[i]]] <- caldf$cv5rmse_13C[i]
-    loocv13c[var_inds_in_calperiod[[i]]] <- caldf$loocv_13C[i]
+   # cv5rmse13c[var_inds_in_calperiod[[i]]] <- caldf$cv5rmse_13C[i]
+  #  cvloo13c[var_inds_in_calperiod[[i]]] <- caldf$cvloo_13C[i]
 
   }
 
@@ -198,12 +198,13 @@ calibrate_ambient_carbon_Bowling2003 <- function(amb_data_list,
   }
 
   # calculate uncertainties:
-  amb_delta$CVcalUcrt <- round(abs(amb_delta$mean_cal) *
-                                sqrt((cv5rmse12c/mean12c)^2 + (cv5rmse13c/mean13c)^2), 3)
-  amb_delta$LOOcalUcrt <- round(abs(amb_delta$mean_cal) *
-                                 sqrt((loocv12c/mean12c)^2 + (loocv13c/mean13c)^2), 3)
-  amb_co2$CVcalUcrt   <- round(sqrt(cv5rmse12c^2 + cv5rmse13c^2), 3)
-  amb_co2$LOOcalUcrt  <- round(sqrt(loocv12c^2 + loocv13c^2), 3)
+  # save this ucrt propogation for later version.
+  #amb_delta$CVcalUcrt <- round(abs(amb_delta$mean_cal) *
+  #                              sqrt((cv5rmse12c/mean12c)^2 + (cv5rmse13c/mean13c)^2), 3)
+  #amb_delta$LOOcalUcrt <- round(abs(amb_delta$mean_cal) *
+  #                               sqrt((cvloo12c/mean12c)^2 + (cvloo13c/mean13c)^2), 3)
+  #amb_co2$CVcalUcrt   <- round(sqrt(cv5rmse12c^2 + cv5rmse13c^2), 3)
+  #amb_co2$LOOcalUcrt  <- round(sqrt(cvloo12c^2 + cvloo13c^2), 3)
 
   # replace ambdf in amb_data_list, return amb_data_list
   amb_data_list$dlta13CCo2 <- amb_delta

diff --git a/R/calibrate_ambient_carbon_linreg.R b/R/calibrate_ambient_carbon_linreg.R
@@ -123,7 +123,16 @@ calibrate_ambient_carbon_linreg <- function(amb_data_list,
     # extract d13C and CO2 concentrations from the ambient data
     d13C_ambdf$mean_cal <- d13C_ambdf$mean
     co2_ambdf$mean_cal  <- co2_ambdf$mean
-
+
+    # add columns to d13C_ambdf and co2_ambdf for uncertainty calculation
+    d13C_ambdf$cvloo   <- d13C_ambdf$mean
+    d13C_ambdf$cv5rmse <- d13C_ambdf$mean
+    d13C_ambdf$cv5mae  <- d13C_ambdf$mean
+
+    co2_ambdf$cvloo    <- co2_ambdf$mean
+    co2_ambdf$cv5rmse  <- co2_ambdf$mean
+    co2_ambdf$cv5mae   <- co2_ambdf$mean
+
     for (i in 1:length(var_inds_in_calperiod)) {
 
       d13C_ambdf$mean_cal[var_inds_in_calperiod[[i]]] <- caldf$d13C_intercept[i] +
@@ -135,6 +144,8 @@ calibrate_ambient_carbon_linreg <- function(amb_data_list,
       d13C_ambdf$max[var_inds_in_calperiod[[i]]]  <- caldf$d13C_intercept[i] +
         d13C_ambdf$max[var_inds_in_calperiod[[i]]] * caldf$d13C_slope[i]
 
+     # d13C_ambdf$cvloo[var_inds_in_calperiod[[i]]] <-  
+
       co2_ambdf$mean_cal[var_inds_in_calperiod[[i]]] <- caldf$co2_intercept[i] +
         co2_ambdf$mean[var_inds_in_calperiod[[i]]] * caldf$co2_slope[i]
 

diff --git a/R/calibrate_carbon.R b/R/calibrate_carbon.R
@@ -153,7 +153,7 @@ calibrate_carbon         <- function(inname,
 
     tmp_names <- names(ciso$reference)
 
-    print("correcting reference output df.,..")
+    print("correcting reference output df...")
     #apply seems to strip names from ciso$reference, so need to save above
     # and reassign below.
     ciso$reference <- lapply(names(ciso$reference),

diff --git a/R/output_functions.R b/R/output_functions.R
@@ -296,9 +296,6 @@ write_carbon_ambient_data <- function(outname, site, amb_data_list) {
     for (i in 1:length(amb_data_list)) {
       amb_data_subset <- amb_data_list[i]
 
-      print(names(amb_data_subset))
-      print(site)
-
       co2_data_outloc <- rhdf5::H5Gcreate(fid,
                                           paste0("/", site, "/dp01/data/isoCo2/", 
                                                  names(amb_data_subset)))

diff --git a/R/reference_data_regression.R b/R/reference_data_regression.R
@@ -104,8 +104,8 @@ fit_carbon_regression <- function(ref_data, method, calibration_half_width,
                         offset13C = as.numeric(NA),
                         r2_12C = as.numeric(NA),
                         r2_13C = as.numeric(NA),
-                        loocv_12C = as.numeric(NA),
-                        loocv_13C = as.numeric(NA),
+                        cvloo_12C = as.numeric(NA),
+                        cvloo_13C = as.numeric(NA),
                         cv5mae_12C = as.numeric(NA),
                         cv5mae_13C = as.numeric(NA),
                         cv5rmse_12C = as.numeric(NA),
@@ -118,8 +118,8 @@ fit_carbon_regression <- function(ref_data, method, calibration_half_width,
                         offset13C = numeric(length = 2e5),
                         r2_12C    = numeric(length = 2e5),
                         r2_13C    = numeric(length = 2e5),
-                        loocv_12C = numeric(length = 2e5),
-                        loocv_13C = numeric(length = 2e5),
+                        cvloo_12C = numeric(length = 2e5),
+                        cvloo_13C = numeric(length = 2e5),
                         cv5mae_12C = numeric(length = 2e5),
                         cv5mae_13C = numeric(length = 2e5),
                         cv5rmse_12C = numeric(length = 2e5),
@@ -183,8 +183,8 @@ fit_carbon_regression <- function(ref_data, method, calibration_half_width,
           out$r2_13C[i] <- summary(tmpmod13C)$r.squared
 
           # extract leave-one-out CV value
-          out$loocv_12C[i] <- loocv(tmpmod12C)
-          out$loocv_13C[i] <- loocv(tmpmod13C)
+          out$cvloo_12C[i] <- loocv(tmpmod12C)
+          out$cvloo_13C[i] <- loocv(tmpmod13C)
 
           # get cv5 values  
           tmp <- stats::formula(conc12CCO2_ref ~ conc12CCO2_obs)
@@ -206,8 +206,8 @@ fit_carbon_regression <- function(ref_data, method, calibration_half_width,
           out$offset13C[i]    <- NA
           out$r2_12C[i]       <- NA
           out$r2_13C[i]       <- NA
-          out$loocv_12C[i]    <- NA
-          out$loocv_13C[i]    <- NA          
+          out$cvloo_12C[i]    <- NA
+          out$cvloo_13C[i]    <- NA          
           out$cv5mae_12C[i]   <- NA
           out$cv5mae_13C[i]   <- NA
           out$cv5rmse_12C[i]  <- NA
@@ -225,9 +225,9 @@ fit_carbon_regression <- function(ref_data, method, calibration_half_width,
       # re-order columns to ensure that they are consistent across methods
       out <- out[, c("timeBgn", "timeEnd",
                      "gain12C", "offset12C", "r2_12C", 
-                     "loocv_12C", "cv5mae_12C", "cv5rmse_12C",
+                     "cvloo_12C", "cv5mae_12C", "cv5rmse_12C",
                      "gain13C", "offset13C", "r2_13C",
-                     "loocv_13C", "cv5mae_13C", "cv5rmse_13C")]
+                     "cvloo_13C", "cv5mae_13C", "cv5rmse_13C")]
 
     }
   } else if (method == "linreg") {
@@ -245,18 +245,30 @@ fit_carbon_regression <- function(ref_data, method, calibration_half_width,
                         d13C_slope     = as.numeric(NA),
                         d13C_intercept = as.numeric(NA),
                         d13C_r2        = as.numeric(NA),
+                        d13C_cvloo     = as.numeric(NA),
+                        d13C_cv5mae    = as.numeric(NA),
+                        d13C_cv5rmse   = as.numeric(NA),
                         co2_slope      = as.numeric(NA),
                         co2_intercept  = as.numeric(NA),
-                        co2_r2         = as.numeric(NA))
+                        co2_r2         = as.numeric(NA),
+                        co2_cvloo      = as.numeric(NA),
+                        co2_cv5mae     = as.numeric(NA),
+                        co2_cv5rmse    = as.numeric(NA))
     } else {
 
       # output dataframe giving valid time range, slopes, intercepts, rsquared.
       out <- data.frame(d13C_slope     = numeric(length = 2e5),
                         d13C_intercept = numeric(length = 2e5),
                         d13C_r2        = numeric(length = 2e5),
+                        d13C_cvloo     = numeric(length = 2e5),
+                        d13C_cv5mae    = numeric(length = 2e5),
+                        d13C_cv5rmse   = numeric(length = 2e5),
                         co2_slope      = numeric(length = 2e5),
                         co2_intercept  = numeric(length = 2e5),
-                        co2_r2         = numeric(length = 2e5))
+                        co2_r2         = numeric(length = 2e5),
+                        co2_cvloo      = numeric(length = 2e5),
+                        co2_cv5mae     = numeric(length = 2e5),
+                        co2_cv5rmse    = numeric(length = 2e5))
 
       # get start and end days.
       start_date <- as.Date(min(ref_data$timeBgn))
@@ -302,6 +314,7 @@ fit_carbon_regression <- function(ref_data, method, calibration_half_width,
           tmpmod_co2 <- stats::lm(rtioMoleDryCo2Refe.mean ~ rtioMoleDryCo2.mean,
                                   data = cal_subset)
 
+
           out$d13C_slope[i]     <- coef(tmpmod_d13)[[2]]
           out$d13C_intercept[i] <- coef(tmpmod_d13)[[1]]
           out$d13C_r2[i]        <- summary(tmpmod_d13)$r.squared
@@ -310,6 +323,23 @@ fit_carbon_regression <- function(ref_data, method, calibration_half_width,
           out$co2_intercept[i]  <- coef(tmpmod_co2)[[1]]
           out$co2_r2[i]         <- summary(tmpmod_co2)$r.squared
 
+          # extract uncertainties:
+          # extract leave-one-out CV value
+          out$d13C_cvloo[i] <- loocv(tmpmod_d13)
+          out$co2_cvloo[i]  <- loocv(tmpmod_co2)
+
+          # get cv5 values  
+          tmp <- stats::formula(dlta13CCo2Refe.mean ~ dlta13CCo2.mean)
+          cv_d13C <- estimate_calibration_error(tmp, cal_subset)
+          tmp <- stats::formula(rtioMoleDryCo2Refe.mean ~ rtioMoleDryCo2.mean)
+          cv_co2 <- estimate_calibration_error(tmp, cal_subset)
+
+          # assign cv values:
+          out$d13C_cv5mae[i]  <- cv_d13C$MAE
+          out$co2_cv5mae[i]   <- cv_co2$MAE
+          out$d13C_cv5rmse[i] <- cv_d13C$RMSE
+          out$co2_cv5rmse[i]  <- cv_co2$RMSE
+
         } else {
 
           out$d13C_slope[i]     <- NA
@@ -320,6 +350,14 @@ fit_carbon_regression <- function(ref_data, method, calibration_half_width,
           out$co2_intercept[i]  <- NA
           out$co2_r2[i]         <- NA
 
+          # set uncertainty values to NA as well.
+          out$d13C_cvloo[i]   <- NA
+          out$d13C_cv5mae[i]  <- NA
+          out$d13C_cv5rmse[i] <- NA
+          out$co2_cvloo[i]    <- NA
+          out$co2_cv5mae[i]   <- NA
+          out$co2_cv5rmse[i]  <- NA
+
         }
       }
 
@@ -333,7 +371,9 @@ fit_carbon_regression <- function(ref_data, method, calibration_half_width,
       # re-order columns to ensure that they are consistent across methods
       out <- out[, c("timeBgn", "timeEnd",
                      "d13C_slope", "d13C_intercept", "d13C_r2",
-                     "co2_slope", "co2_intercept", "co2_r2")]
+                     "d13C_cvloo", "d13C_cv5mae", "d13C_cv5rmse",
+                     "co2_slope", "co2_intercept", "co2_r2",
+                     "co2_cvloo", "co2_cv5mae", "co2_cv5rmse")]
 
     }
   }

diff --git a/R/restructure_data.R b/R/restructure_data.R
@@ -88,9 +88,6 @@ ingest_data <- function(inname, analyte, name_fix = TRUE) {
 
   ambi_by_height <- base::split(ambient, factor(ambient$verticalPosition)) 
   refe_by_height <- base::split(reference, factor(reference$verticalPosition))
-
-  print(names(ambi_by_height))
-  print(names(refe_by_height))
 
   #-------------------------
   # RESTRUCTURE AMBIENT
@@ -144,9 +141,6 @@ ingest_data <- function(inname, analyte, name_fix = TRUE) {
     names(refe_out) <- paste0(names(refe_out), "_09m")
   }
 
-  print(names(refe_out))
-  print(names(ambi_out))
-
   output <- list(ambi_out, refe_out, reference)
   names(output) <- c("ambient", "reference", "refe_stacked")
 

diff --git a/cran-comments.md b/cran-comments.md
@@ -1,19 +1,29 @@
 ## Test environments
-* local R installation: R 4.1.2
+* local R installation: R 4.2.1
 * GitHub Actions (ubuntu-18.04): devel, release, oldrel
 * GitHub Actions (windows-latest): release
 * GitHub Actions (macOS-latest): devel, release, oldrel
-* win-builder: devel
+* win-builder: devel, release
+* R-hub (Windows Server 2022): devel
+* R-hub (ubuntu-20.04.1 LTS): release
+* R-hub (fedora): devel
 
 ## R CMD check results
 
 * No errors, warnings, or notes in local environment.
-* No errors, warnings, or notes in GitHub Actions, except for windows-latest where
-R CMD check does not run due to an error in 'session info' 
-(Error: invalid version specification '<NA'>). This error does not seem to be related
-to the package itself.
+
+* No errors, warnings, or notes in GitHub Actions, except for macOS-latest (devel),
+which failed to download R-devel due to a 404 HTTP error
+
 * No errors or warnings on win-builder.
 
+* One error and one note issued on R-hub/Windows Server (devel): error is about a 
+bioconductor dependency not being available, note is about change in maintainer
+(my email address has changed) and invalid URLs (URLs are correct however). 
+Same note is also provided on ubuntu-20.04.1 LTS.
+
+* No errors, warnings, or notes on R-hub fedora.
+
 * Auto-check will issue a warning about misspelled words in DESCRIPTION, but the words flagged are all correct.
 
 

diff --git a/extract_data_for_Gabe.R b/extract_data_for_Gabe.R
diff --git a/man/calibrate_ambient_carbon_Bowling2003.Rd b/man/calibrate_ambient_carbon_Bowling2003.Rd
diff --git a/setup_testing_data.R b/setup_testing_data.R
@@ -8,9 +8,9 @@ library(rhdf5)
 # exists in rbuildignore, so should not be bundled with package.
 # requires install and restart w/ NEONiso.
 
-master_file <- "/Volumes/GradSchoolBackup/NEON/DP4_00200_001/ONAQ/NEON.D15.ONAQ.DP4.00200.001.nsae.2019-06.basic.h5"
-master_ccalB03_file <- "~/Desktop/NEONcal/210303_CisoEC/ONAQ/NEON.D15.ONAQ.DP4.00200.001.nsae.2019-06.basic.20201020T212232Z.calibrated.h5"
-master_ccalLR_file <- "~/Desktop/NEONcal/210303_CisoLR/ONAQ/NEON.D15.ONAQ.DP4.00200.001.nsae.2019-06.basic.20201020T212232Z.calibrated.h5"
+master_file <- "NEON/DP4_00200_001/ONAQ/NEON.D15.ONAQ.DP4.00200.001.nsae.2019-06.basic.h5"
+master_ccalB03_file <- "210303_CisoEC/ONAQ/NEON.D15.ONAQ.DP4.00200.001.nsae.2019-06.basic.20201020T212232Z.calibrated.h5"
+master_ccalLR_file <- "210303_CisoLR/ONAQ/NEON.D15.ONAQ.DP4.00200.001.nsae.2019-06.basic.20201020T212232Z.calibrated.h5"
 
 #---------------------------------------
 # extract co2 data for unit testing:
@@ -462,5 +462,5 @@ h5delete(fid, '/YELL/dp01/data/isoCo2/co2High_30m')
 
 H5Fclose(fid)
 
-system('/Users/rfiorella/opt/anaconda3/bin/h5repack inst/extdata/NEON.D12.YELL.DP4.00200.001.nsae.2020-11.basic.20210209T161116Z.h5 inst/extdata/NEON.D12.YELL.DP4.00200.001.nsae.2020-11.basic.packed.h5')
+system('~/opt/anaconda3/bin/h5repack inst/extdata/NEON.D12.YELL.DP4.00200.001.nsae.2020-11.basic.20210209T161116Z.h5 inst/extdata/NEON.D12.YELL.DP4.00200.001.nsae.2020-11.basic.packed.h5')