Fix warnings when calculating circular stats on NA sets

R/generic_stats.R

@@ -305,15 +305,18 @@ compute_circular_mean <- function(direction_degrees, date.factors, format) {
   direction_degrees <- direction_degrees[valid_idx]
   date.factors <- date.factors[valid_idx]
 
-  if (length(direction_degrees) == 0) {
-    return(NA)  # If all are NA, return NA
-  }
-
   # Convert directions to 'circular' objects
   directions_circular <- circular::circular(direction_degrees, units = "degrees", modulo = "2pi")
 
   # Compute circular mean
-  circular_mean <- tapply(directions_circular, date.factors, circular::mean.circular, na.rm = TRUE)
+  circular_mean <- tapply(directions_circular, date.factors, function(x) {
+    if (all(is.na(x))) {
+      return(NA)  # Return NA if the entire group is NA
+    } else {
+      return(circular::mean.circular(x, na.rm = TRUE))
+    }
+  })
+
 
   # Convert back to degrees
   circular_mean_degrees <- as.numeric(circular_mean)
@@ -354,7 +357,14 @@ compute_circular_sd <- function(direction_degrees, date.factors) {
   directions_circular <- circular::circular(direction_degrees, units = "degrees", modulo = "2pi")
 
   # Compute circular standard deviation
-  circular_sd <- tapply(directions_circular, date.factors, circular::sd.circular, na.rm = TRUE)
+  circular_sd <- tapply(directions_circular, date.factors, function(x) {
+    if (all(is.na(x))) {
+      return(NA)  # Return NA if the entire group is NA
+    } else {
+      return(circular::sd.circular(x, na.rm = TRUE))
+    }
+  })
+
   circular_sd_degrees <- as.numeric(circular_sd) * (180 / pi)  # Convert from radians to degrees
   circular_sd_degrees[is.nan(circular_sd_degrees)] <- NA
   return(circular_sd_degrees)

tests/test_generic_scalar_stats.R

@@ -62,15 +62,13 @@ climdex.pcic.test.scalar.exact.dates <- function() {
     northern.hemisphere = TRUE,
     calendar = "gregorian"
   )
-  # Monthly max
-  result <- compute.stat.scalar(scalar_obj, stat = "max", freq = "monthly", include.exact.dates = TRUE)
-
+  # Annual max
+  result <- compute.stat.scalar(scalar_obj, stat = "max", freq = "annual", include.exact.dates = TRUE)
   expected_max <- max(data)
   max_index <- which.max(data)
   expected_max_date <- dates[max_index]
-
   computed_exact_date <- as.PCICt(result$ymd[1], cal = "gregorian")
-  
+
   checkEqualsNumeric(as.numeric(result$val[1]), expected_max, 
                      msg = paste("Computed max statistic:", result$val[1], 
                                  "Expected max statistic:", expected_max))
@@ -117,7 +115,6 @@ climdex.pcic.test.compute.stat.scalar.sd <- function() {
   # Monthly SD
   result <- compute.stat.scalar(scalar_obj, stat = "sd", freq = "monthly", include.exact.dates = FALSE)
   expected_sd <- sd(data)
-
   checkEqualsNumeric(as.numeric(result[1]), expected_sd, tolerance = 1e-6, 
                      msg = paste("Computed SD:", result[1], "Expected SD:", expected_sd))
 }

tests/test_generic_vector_stats.R

@@ -304,7 +304,6 @@ climdex.pcic.test.compute.stat.vector.filtered.direction.crossing.360 <- functio
     include.exact.dates = FALSE,
     direction.range = c(350, 10)
   )
-  result
   filtered_speeds <- speed[direction >= 350 | direction <= 10]
   expected_max_magnitude <- max(filtered_speeds)
   checkEqualsNumeric(as.numeric(result$magnitude[1]), expected_max_magnitude)
@@ -334,7 +333,7 @@ climdex.pcic.test.compute.stat.vector.no.data.in.direction.range <- function() {
     include.exact.dates = FALSE,
     direction.range = c(270, 360)
   )
-  result
+
   result$magnitude[1]
   checkTrue(is.na(result$magnitude[1]), "Expected NA result when no data is in the specified direction range.")
 }