Add RMSE to output diagnostics (#73)

Author: bpbond

NAMESPACE

@@ -31,6 +31,7 @@ importFrom(lubridate,ymd_hms)
 importFrom(stats,approx)
 importFrom(stats,lm)
 importFrom(stats,na.omit)
+importFrom(stats,predict)
 importFrom(utils,browseURL)
 importFrom(utils,head)
 importFrom(utils,read.csv)

R/models.R

@@ -14,8 +14,8 @@
 #' diffusion theory; see Hutchinson and Mosier (1981) and Nakano et al. (2004).
 #'
 #' For each model type, the following columns are returned:
-#' * Model statistics \code{AIC}, \code{r.squared}, \code{sigma},
-#'  and \code{p.value};
+#' * Model statistics \code{AIC}, \code{r.squared}, \code{RMSE},
+#' and \code{p.value};
 #' * Flux (slope) statistics \code{flux.estimate} and \code{flux.std.error};
 #' * Intercept statistics \code{int.estimate} and \code{int.std.error};
 #' * For the robust linear regression model only,
@@ -38,7 +38,7 @@
 #' 1981. \url{http://dx.doi.org/10.2136/sssaj1981.03615995004500020017x}
 #' @importFrom broom glance tidy
 #' @importFrom MASS rlm
-#' @importFrom stats lm
+#' @importFrom stats lm predict
 #' @export
 #' @examples
 #' # Toy data - linear
@@ -64,9 +64,10 @@ ffi_fit_models <- function(time, conc, area, volume) {
   # Linear model overall metrics. 'glance' produces 12 different ones;
   # we keep the first 5 (adjR2, R2, sigma, statistic, p-value)
   lin_model_stats <- glance(mod)[c("r.squared", "sigma", "p.value", "AIC")]
+
   names(lin_model_stats) <- paste0("lin_", names(lin_model_stats))
 
-  # Slope and intercept statistics
+    # Slope and intercept statistics
   tmod <- tidy(mod)
   lin_slope_stats <- tmod[2, c("estimate", "std.error")]
   names(lin_slope_stats) <- paste0("lin_flux.", names(lin_slope_stats))
@@ -84,11 +85,13 @@ ffi_fit_models <- function(time, conc, area, volume) {
     rob_model_stats <- glance(robust)[c("sigma", "converged", "AIC")]
     tmod <- tidy(robust)
     rob_slope_stats <- tmod[2, c("estimate", "std.error")]
+
 #    rob_int_stats <- tmod[1, c("estimate", "std.error")]
   },
   error = function(e) {
     warning("Could not fit robust linear model")
-    rob_model_stats <- data.frame(sigma = NA_real_, converged = FALSE, AIC = NA_real_)
+    rob_model_stats <- data.frame(sigma = NA_real_, converged = FALSE,
+                                  AIC = NA_real_, RMSE = NA_real_)
     rob_slope_stats <- data.frame(estimate = NA_real_, std.error = NA_real_)
     # rob_int_stats <- data.frame(estimate = NA_real_, std.error = NA_real_)
   })
@@ -101,11 +104,11 @@ ffi_fit_models <- function(time, conc, area, volume) {
   # int_stats <- cbind(int_stats, rob_int_stats)
 
   # Add polynomial regression as a QA/QC check
-  poly_model_stats <- data.frame(r.squared = NA_real_, AIC = NA_real_)
+  poly_model_stats <- data.frame(r.squared = NA_real_, AIC = NA_real_, RMSE = NA_real_)
   if(length(time) > 3) {
     try({
       poly <- lm(conc ~ poly(time, 3))
-      poly_model_stats <- glance(poly)[c("r.squared", "AIC")]
+      poly_model_stats <- glance(poly)[c("r.squared", "sigma", "AIC")]
     })
   }
 
@@ -117,11 +120,11 @@ ffi_fit_models <- function(time, conc, area, volume) {
 
   # The HM1981 approach is based on an exponential model, so derive fit
   # statistics by log-transforming the data
-  if(!is.na(slope_stats$HM81_flux.estimate)) {
-    ffi_message("NOTE: HM81_flux.estimate is not NA, implying nonlinear data")
+  if(is.na(slope_stats$HM81_flux.estimate)) {
     hm81_model_stats <- data.frame(r.squared = NA_real_, sigma = NA_real_,
                                    p.value = NA_real_, AIC = NA_real_)
   } else {
+    ffi_message("NOTE: HM81_flux.estimate is not NA, implying nonlinear data")
     # The time values are probably normalized, i.e. starting at zero
     # Add a (presumably) tiny offset so we don't get log(0) errors
     mod <- lm(conc ~ log(time + 0.01))
@@ -131,6 +134,9 @@ ffi_fit_models <- function(time, conc, area, volume) {
   names(hm81_model_stats) <- paste0("HM81_", names(hm81_model_stats))
   model_stats <- cbind(model_stats, hm81_model_stats)
 
+  # Change "sigma" to "RMSE"; more intuitive for most users
+  names(model_stats) <- gsub("sigma", "RMSE", names(model_stats))
+
   # Combine, sort columns, and return
   out <- cbind(model_stats, slope_stats, int_stats)
   return(out[sort(names(out))])

man/ffi_fit_models.Rd

@@ -275,8 +275,9 @@ x <- round(x, 3)
 From the diagnostics returned by `ffi_compute_fluxes`:
 
 * `HM81_flux.estimate` is not `NA`, which only occurs with saturating behavior;
-* The `lin_AIC` (`r x$lin_AIC`) and `rob_AIC` (`r x$rob_AIC`) values are similar, so no indication of influential outliers;
+* The `lin_AIC` (`r x$lin_AIC`) and `rob_AIC` (`r x$rob_AIC`) [Akaike information criterion](https://en.wikipedia.org/wiki/Akaike_information_criterion) values are similar, so no indication of influential outliers;
 * The `lin_r.squared` (`r x$lin_r.squared`) and `poly_r.squared` (`r x$poly_r.squared`) values are _very_ different, suggesting a failure of the linear model;
+* The [root mean square error](https://en.wikipedia.org/wiki/Root_mean_square_deviation) (RMSE) of the linear model is much higher than the other models' values;
 * The `HM81_r.squared` (`r x$HM81_r.squared`) and `HM81_AIC` (`r x$HM81_AIC`) are considerably higher and lower, respectively, than the linear model. 
 
 All of these metrics point to a common conclusion: a linear model is _not_