DrillDown into Random Parameter & Program & Site

A drilldown into a set of random parameter+program+site combinations.

library(here)
library(dplyr)
source(here("SEACARProgramCompare/mapProgramNameToShortName.R"))


# === new dataset
df_SEACAR <- readr::read_delim(
  here("data/Discrete WQ - 10006.txt"),
  delim = "|"
) %>%
  # align programName between datasets
  mutate(ProgramName = mapProgramNameToShortName(ProgramName))



# === previous year's dataset
df_OLD <- readr::read_delim(here::here("data/allDataSEACAR.csv"), delim=",") %>%
  # align programName between datasets
  mutate(ProgramName = mapProgramNameToShortName(ProgramName))



# === keep only columns of interest to avoid coltype errors
cols_of_interest <- c(
  "ProgramName", 
  "ParameterName",
  "ResultValue", 
  "SampleDate",
  "ProgramLocationID"
)
df_SEACAR <- df_SEACAR %>% select(all_of(cols_of_interest))
df_OLD <- df_OLD %>% select(all_of(cols_of_interest))


# === Combine dataframes with a source column
df_combined <- bind_rows(
  df_SEACAR %>% mutate(source = "SEACAR_STD"),
  df_OLD %>% mutate(source = "OLD_STD")
)

Random Sample Comparisons

The following is visualization of a random sample of program+parameter combinations.

library(ggplot2)
# === 3 parameters from one program
random_program <- sample(unique(df_combined$ProgramName), 1)
program_parameters <- unique(df_combined$ParameterName[df_combined$ProgramName == random_program])
random_parameters <- sample(program_parameters, min(3, length(program_parameters)))

# Select a random location ID from the program
program_locations <- unique(df_combined$ProgramLocationID[df_combined$ProgramName == random_program])
random_location <- sample(program_locations, 1)

cat("=== Program:", random_program, "===\n")

=== Program: BROWARD ===

cat("Location:", random_location, "===\n")

Location: 18 ===

for (i in 1:length(random_parameters)) {
  random_parameter <- random_parameters[i]
  
  df_subset <- df_combined %>%
    filter(ProgramName == random_program, ParameterName == random_parameter, ProgramLocationID == random_location)
  
  cat("Parameter:", random_parameter, "\n")
  
  # === violin plot
  p1 <- ggplot(df_subset, aes(x = source, y = ResultValue, fill = source)) +
    geom_violin(alpha = 0.6) +
    geom_boxplot(width = 0.2, alpha = 0.8, outlier.shape = NA) +
    scale_y_log10() +
    labs(
      title = paste("ResultValue Distribution for", random_program, "-", random_location, "-", random_parameter),
      x = "Dataset",
      y = "ResultValue (log10)"
    ) +
    theme_minimal() +
    theme(legend.position = "none")
  
  print(p1)
  
  # === time series plot
  p2 <- ggplot(df_subset, aes(x = SampleDate, y = ResultValue, color = source)) +
    geom_point(shape = 4, alpha = 0.4) +
    scale_y_log10() +
    labs(
      title = paste("ResultValue Over Time for", random_program, "-", random_location, "-", random_parameter),
      x = "Sample Date",
      y = "ResultValue"
    ) +
    theme_minimal() +
    theme(legend.position = "bottom")
  
  print(p2)
}

Parameter: Salinity 

Parameter: Dissolved Oxygen Saturation 

Parameter: Dissolved Oxygen