This notebook explores the results from the scripts in the folder calibration_scenarios_Xweeks_Xmonths. Note that these scripts require a lot of computational time and a lost of space to store the results of the different calibration (about 174Gb). For each calibration scenario, the script has been divided into several files so they can be ran in parallel (for instance using background jobs, launched from the root directory) to speed up the calculations.

These represents the different calibration scenarios presented in the paper:

1 week of pre-deployment calibration, 2 months of evaluation, 1 week of post-deployment calibration
2 weeks of pre-deployment calibration, 2 months of evaluation, 2 weeks of post-deployment calibration
2 weeks of pre-deployment calibration, 4 months of evaluation, 2 weeks of post-deployment calibration
2 weeks of pre-deployment calibration, 6 months of evaluation, 2 weeks of post-deployment calibration
1 week of pre-deployment calibration, 2 months of evaluation
2 weeks of pre-deployment calibration, 2 months of evaluation
4 weeks of pre-deployment calibration, 2 months of evaluation

source("utilities/utilities.R")

## Warning: package 'openair' was built under R version 4.1.2

source("utilities/nested_models.r")
source("utilities/variables.r")

require(tidytext)

## Warning: package 'tidytext' was built under R version 4.1.3

Data preparation

input_folder <-
  paste0(here::here(),
         "/output/calibration/scenarios/calibration_scenarios_comparison/")

res <-
  readRDS(paste0(input_folder, "res_calibration_scenario_comparison.rds"))
res_bus <-
  readRDS(paste0(input_folder, "res_bus_calibration_scenario_comparison.rds"))

res$method <-
  fct_recode(
    res$method,
    `Koehler_mass+LR` = "koehlermasslr",
    `Koehler_mass+OLS` = "koehlermassols",
    `Koehler_size+LR` = "koehlersizelr",
    `Koehler_size+OLS` =  "koehlersizeols",
    `Laulainen+LR` = "laulainenlr",
    `Laulainen+OLS` = "laulainenols",
    `MLR_part.` =  "lmpart",
    `MLR_part.+RH` =   "lmpartrh",
    `LR` = "lr",
    `MLR_RH` = "mlr1" ,
    `OLS` =  "ols",
    `RLM` = "rlmhuber",
    `RLM_part.+RH`  = "rlmpartrh",
    `RLM+RH` =      "rlmrh",
    `RLM_part.` = "rlmpart"
  )
res$method <-
  fct_relevel(
    res$method,
    "Initial",
    "OLS",
    "LR",
    "MLR_RH",
    "MLR_part.",
    "MLR_part.+RH",
    "RLM",
    "RLM+RH",
    "RLM_part.",
    "RLM_part.+RH",
    "Koehler_mass+LR",
    "Koehler_mass+OLS",
    "Koehler_size+LR",
    "Koehler_size+OLS",
    "Laulainen+LR",
    "Laulainen+OLS"
  )

res_bus$method <-
  fct_recode(
    res_bus$method,
    `Koehler_mass+LR` = "koehlermasslr",
    `Koehler_mass+OLS` = "koehlermassols",
    `Koehler_size+LR` = "koehlersizelr",
    `Koehler_size+OLS` =  "koehlersizeols",
    `Laulainen+LR` = "laulainenlr",
    `Laulainen+OLS` = "laulainenols",
    `MLR_part.` =  "lmpart",
    `MLR_part.+RH` =   "lmpartrh",
    `LR` = "lr",
    `MLR_RH` = "mlr1" ,
    `OLS` =  "ols",
    `RLM` = "rlmhuber",
    `RLM_part.+RH`  = "rlmpartrh",
    `RLM+RH` =      "rlmrh",
    `RLM_part.` = "rlmpart"
  )
res_bus$method <-
  fct_relevel(
    res_bus$method,
    "Initial",
    "OLS",
    "LR",
    "MLR_RH",
    "MLR_part.",
    "MLR_part.+RH",
    "RLM",
    "RLM+RH",
    "RLM_part.",
    "RLM_part.+RH",
    "Koehler_mass+LR",
    "Koehler_mass+OLS",
    "Koehler_size+LR",
    "Koehler_size+OLS",
    "Laulainen+LR",
    "Laulainen+OLS"
  )

res$scenario_type <-
  fct_recode(
    res$scenario_type,
    `1w8w1w` = "1w2m1w",
    `2w8w` = "2w",
    `2w8w2w` = "2w2m2w",
    `4w8w` =  "4w",
    `2w16w2w` = "2w4m2w"
  )

res$scenario_type <-
  fct_relevel(res$scenario_type,
              "2w8w",
              "4w8w",
              "1w8w1w",
              "2w8w2w",
              "2w16w2w",
              "2w24w2w")

res_bus$scenario_type <-
  fct_recode(
    res_bus$scenario_type,
    `1w8w1w` = "1w2m1w",
    `2w8w` = "2w",
    `2w8w2w` = "2w2m2w",
    `4w8w` =  "4w" ,
    `2w16w2w` = "2w4m2w"
  )

res_bus$scenario_type <-
  fct_relevel(res_bus$scenario_type,
              "2w8w",
              "4w8w",
              "1w8w1w",
              "2w8w2w",
              "2w16w2w",
              "2w24w2w")

res <- res[!grepl("N2", res$sensor),]
res_bus <- res_bus[!grepl("N2", res_bus$combination),]

Filter out techniques discarded by robust selection

res <-
  res[!(
    res$method %in% c(
      "Koehler_mass+LR",
      "Koehler_mass+OLS",
      "Koehler_size+LR",
      "Koehler_size+OLS",
      "Laulainen+LR",
      "Laulainen+OLS",
      "koehlermass",
      "koehlersize",
      "laulainen"
    )
  ),] %>%
  dplyr::mutate(sensor_type = ifelse(grepl("SPS", sensor), "SPS30", "PMS5003"))

res_bus <-
  res_bus[!(
    res_bus$method %in% c(
      "Koehler_mass+LR",
      "Koehler_mass+OLS",
      "Koehler_size+LR",
      "Koehler_size+OLS",
      "Laulainen+LR",
      "Laulainen+OLS",
      "koehlermass",
      "koehlersize",
      "laulainen"
    )
  ),] %>%
  dplyr::mutate(sensor_type = ifelse(grepl("SPS", combination), "SPS30", "PMS5003"))

Results

RMSE and Between Unit Uncertainty

p_rmse <- res %>%
  dplyr::filter(type == "Full") %>%
  dplyr::filter(dataset_type == "cal") %>%
  ggplot(aes(x = method, y = RMSE, colour = scenario_type)) +
  stat_boxplot(geom = "errorbar") +
  geom_boxplot(
    fatten = 0.8,
    outlier.size = 0.3,
    outlier.alpha = 0.5,
    lwd = 0.2
  ) +
  custom_theme +
  theme(axis.text.x = element_text(
    angle = 90,
    hjust = 0.95,
    vjust = 0.2
  )) +
  facet_wrap( ~ sensor_type, ncol = 2, scales = "free_y") +
  xlab("Calibration method") +
  ylab(quickText("RMSE (ug/m3)")) +
  ylim(c(0, quantile(res$RMSE, probs = 0.9)))
#p_rmse




p_bus <- res_bus %>%
  dplyr::filter(type == "Full") %>%
  dplyr::filter(dataset_type == "cal") %>%
  ggplot(aes(
    x = method,
    y = usb,
    colour = scenario_type
  )) +
  stat_boxplot(geom = "errorbar") +
  geom_boxplot(
    fatten = 0.8,
    outlier.size = 0.3,
    outlier.alpha = 0.5,
    lwd = 0.2
  ) +
  #geom_line(data = table_res, aes(x=method,y=mean_RMSE, colour=sensor))+
  # ggtitle("Calibration - PMS - Scenarios comparison") +
  custom_theme +
  theme(axis.text.x = element_text(
    angle = 90,
    hjust = 0.95,
    vjust = 0.2
  )) +
  facet_wrap( ~ sensor_type, ncol = 2, scales = "free_y") +
  xlab("Calibration method") +
  ylab(quickText("Between unit uncertainty (ug/m3)"))
#p_bus


legend <- get_legend(# create some space to the left of the legend
  p_rmse +
    guides(color = guide_legend(ncol = 1, title = "Sensors")) +
    #theme(legend.position = "bottom") +
    theme(legend.box.margin = margin(0, 0, 0, 0)))

## Warning: Removed 14219 rows containing non-finite values (stat_boxplot).

## Warning: Removed 14219 rows containing non-finite values (stat_boxplot).

pg <-
  plot_grid(
    p_rmse + xlab(NULL) + guides(x = "none") + theme(legend.position = "none"),
    p_bus + theme(legend.position = "none"),
    nrow = 2,
    labels = c("(a)", "(b)"),
    label_y = c(0.1, 0.1)
  )

## Warning: Removed 14219 rows containing non-finite values (stat_boxplot).

## Warning: Removed 14219 rows containing non-finite values (stat_boxplot).

pg_l <- plot_grid(pg,
                  legend,
                  ncol = 2,
                  rel_widths  = c(0.8, 0.2))
pg_l

#ggsave2(plot = pg_l,filename=paste0(output_plot_folder, "comparison_scenarios.svg"),
#    height = 140, units = "mm")

On RLM_part.+RH only

In this section we only focus on the four methods that performed both during the robust method selection ((presented in in Calibration 2 weeks 40 days), with a special focus on RLM_part.+RH for clarity in the graph.

p_rmse <- res %>%
  dplyr::filter(type == "Full") %>%
  dplyr::filter(dataset_type == "cal") %>%
  dplyr::filter(method %in% c("RLM_part.+RH", "RLM_part.", "MLR_part.", "MLR_part.+RH")) %>%
  ggplot(aes(
    x = reorder_within(scenario_type, RMSE, sensor_type, median),
    y = RMSE,
    colour = method
  )) +
  stat_boxplot(geom = "errorbar") +
  geom_boxplot(
    fatten = 0.8,
    outlier.size = 0.3,
    outlier.alpha = 0.5,
    lwd = 0.2
  ) +
  #geom_line(data = table_res, aes(x=method,y=mean_RMSE, colour=sensor))+
  # ggtitle("Calibration - PMS - Scenarios comparison") +
  custom_theme +
  theme(axis.text.x = element_text(
    angle = 90,
    hjust = 0.95,
    vjust = 0.2
  )) +
  facet_wrap( ~ sensor_type, ncol = 2, scales = "free") +
  xlab("Calibration method") +
  ylab(quickText("RMSE (ug/m3)")) +
  ylim(c(0, quantile(res$RMSE, probs = 0.9)))
p_rmse

## Warning: Removed 2662 rows containing non-finite values (stat_boxplot).

## Warning: Removed 2662 rows containing non-finite values (stat_boxplot).

p_rmse <- res %>%
  dplyr::filter(type == "Full") %>%
  dplyr::filter(dataset_type == "cal") %>%
  dplyr::filter(method %in% c("RLM_part.+RH", "RLM_part.", "MLR_part.", "MLR_part.+RH")) %>%
  ggplot(aes(
    x = reorder_within(method, RMSE, sensor_type, median),
    y = RMSE,
    colour = scenario_type
  )) +
  stat_boxplot(geom = "errorbar") +
  geom_boxplot(
    fatten = 0.8,
    outlier.size = 0.3,
    outlier.alpha = 0.5,
    lwd = 0.2
  ) +
  #geom_line(data = table_res, aes(x=method,y=mean_RMSE, colour=sensor))+
  # ggtitle("Calibration - PMS - Scenarios comparison") +
  custom_theme +
  theme(axis.text.x = element_text(
    angle = 90,
    hjust = 0.95,
    vjust = 0.2
  )) +
  facet_wrap( ~ sensor_type, ncol = 2, scales = "free") +
  xlab("Calibration method") +
  ylab(quickText("RMSE (ug/m3)")) +
  ylim(c(0, quantile(res$RMSE, probs = 0.9)))
p_rmse

## Warning: Removed 2662 rows containing non-finite values (stat_boxplot).

## Warning: Removed 2662 rows containing non-finite values (stat_boxplot).

median_IQR <- function(x) {
  data.frame(y = quantile(x)[4],
             label = round2(quantile(x)[4] - quantile(x)[2], n = 1))
}


p_rmse <- res %>%
  dplyr::filter(type == "Full") %>%
  dplyr::filter(dataset_type == "cal") %>%
  dplyr::filter(method %in% c("RLM_part.+RH")) %>%
  ggplot(aes(x = sensor, y = RMSE, colour = scenario_type)) +
  stat_boxplot(geom = "errorbar") +
  geom_boxplot(
    fatten = 1,
    outlier.size = 0.3,
    outlier.alpha = 0.5,
    lwd = 0.2
  ) +
  #geom_line(data = table_res, aes(x=method,y=mean_RMSE, colour=sensor))+
  # ggtitle("Calibration - PMS - Scenarios comparison") +
  custom_theme +
  # theme(axis.text.x = element_text(angle = 90, hjust=0.95,vjust=0.2)) +
  facet_wrap( ~ sensor_type, nrow = 2, scales = "free") +
  xlab("Sensors") +
  ylab(quickText("RMSE (ug/m3)")) +
  #ylim(c(0,quantile(res$RMSE,probs = 0.9))) +
  stat_summary(
    geom = "text",
    fun.data = median_IQR,
    aes(group = scenario_type),
    size = 2,
    position = position_dodge(width = 0.75),
    vjust = -0.5,
    colour = "black"
  ) +
  theme(legend.position = "bottom") + guides(colour = guide_legend(nrow = 1)) +
  scale_y_continuous(limits = c(0, quantile(res$RMSE, probs = 0.9)),
                     minor_breaks = seq(0, 10, 0.2)) +
  labs(colour = "Scenarios")
p_rmse

## Warning: Removed 752 rows containing non-finite values (stat_boxplot).

## Warning: Removed 752 rows containing non-finite values (stat_boxplot).

## Warning: Removed 752 rows containing non-finite values (stat_summary).

ggsave2(
  plot = p_rmse,
  filename = paste0(
    PLOT_OUTPUT_FOLDER,
    "comparison_scenarios_rmse_RLMparRH.png"
  ),
  height = 200,
  units = "mm"
)

## Saving 178 x 200 mm image

## Warning: Removed 752 rows containing non-finite values (stat_boxplot).

## Warning: Removed 752 rows containing non-finite values (stat_boxplot).

## Warning: Removed 752 rows containing non-finite values (stat_summary).

Between unit uncertainty

duplicated_cbn <-
  c(
    "PMS-63N3 - PMS-86N2",
    "PMS-65N4 - PMS-86N2",
    "PMS-75N3 - PMS-86N2",
    "PMS-91N4 - PMS-86N2",
    "SPS-19N4 - SPS-38N2",
    "SPS-21N4 - SPS-38N2",
    "SPS-36N3 - SPS-38N2",
    "SPS-41N3 - SPS-38N2"
  )

p_bus <- res_bus %>%
  dplyr::filter(type == "Full") %>%
  dplyr::filter(dataset_type == "cal") %>%
  dplyr::filter(method %in% c("RLM_part.+RH")) %>%
  dplyr::filter(!(combination %in% duplicated_cbn)) %>%
  ggplot(aes(
    x = str_remove_all(combination, "SPS-|PMS-") ,
    y = usb,
    colour = scenario_type
  )) +
  stat_boxplot(geom = "errorbar") +
  geom_boxplot(
    fatten = 0.8,
    outlier.size = 0.3,
    outlier.alpha = 0.5,
    lwd = 0.2
  ) +
  #geom_line(data = table_res, aes(x=method,y=mean_RMSE, colour=sensor))+
  # ggtitle("Calibration - PMS - Scenarios comparison") +
  custom_theme +
  theme(axis.text.x = element_text(
    angle = 90,
    hjust = 0.95,
    vjust = 0.2
  )) +
  facet_wrap( ~ sensor_type, nrow = 2, scales = "free") +
  xlab("Calibration method") +
  ylab(quickText("Between unit uncertainty (ug/m3)")) +
  stat_summary(
    geom = "text",
    fun =  ~ quantile(.x, probs = c(1, 2, 3) / 4),
    aes(group = scenario_type, label = sprintf("%1.1f", ..y..)),
    size = 2,
    position = position_dodge(width = 0.75),
    vjust = -0.5,
    colour = "black"
  ) +
  theme(legend.position = "bottom") + guides(colour = guide_legend(nrow = 1))
p_bus

p_bus <- res_bus %>%
  dplyr::filter(type == "Full") %>%
  dplyr::filter(dataset_type == "cal") %>%
  dplyr::filter(method %in% c("RLM_part.+RH")) %>%
  ggplot(aes(x = scenario_type , y = usb, colour = scenario_type)) +
  stat_boxplot(geom = "errorbar") +
  geom_boxplot(
    fatten = 0.8,
    outlier.size = 0.3,
    outlier.alpha = 0.5,
    lwd = 0.2
  ) +
  #geom_line(data = table_res, aes(x=method,y=mean_RMSE, colour=sensor))+
  # ggtitle("Calibration - PMS - Scenarios comparison") +
  custom_theme +
  theme(axis.text.x = element_text(
    angle = 90,
    hjust = 0.95,
    vjust = 0.2
  )) +
  facet_wrap( ~ sensor_type, nrow = 2, scales = "free") +
  xlab("Calibration method") +
  ylab(quickText("Between unit uncertainty (ug/m3)")) +
  stat_summary(
    geom = "text",
    fun =  ~ quantile(.x, probs = c(1, 2, 3) / 4),
    aes(group = scenario_type, label = sprintf("%1.1f", ..y..)),
    size = 2,
    position = position_dodge(width = 0.75),
    vjust = -0.5,
    colour = "black"
  )
p_bus

Other metrics

p_bias <- res %>%
  dplyr::filter(type == "Full") %>%
  dplyr::filter(dataset_type == "cal") %>%
  ggplot(aes(x = method, y = y__, colour = scenario_type)) +
  stat_boxplot(geom = "errorbar") +
  geom_boxplot(
    fatten = 0.8,
    outlier.size = 0.3,
    outlier.alpha = 0.5,
    lwd = 0.2
  ) +
  #geom_line(data = table_res, aes(x=method,y=mean_RMSE, colour=sensor))+
  # ggtitle("Calibration - PMS - Scenarios comparison") +
  custom_theme +
  theme(axis.text.x = element_text(
    angle = 90,
    hjust = 0.95,
    vjust = 0.2
  )) +
  facet_wrap( ~ sensor_type, ncol = 2, scales = "free_y") +
  xlab("Calibration method") +
  ylab(quickText("Bias (ug/m3)")) +
  ylim(c(0, quantile(res$y__, probs = 0.9)))
p_bias

## Warning: Removed 98020 rows containing non-finite values (stat_boxplot).

## Warning: Removed 98020 rows containing non-finite values (stat_boxplot).

p_U <- res %>%
  dplyr::filter(type == "Full") %>%
  dplyr::filter(dataset_type == "cal") %>%
  ggplot(aes(x = method, y = U, colour = scenario_type)) +
  stat_boxplot(geom = "errorbar") +
  geom_boxplot(
    fatten = 0.8,
    outlier.size = 0.3,
    outlier.alpha = 0.5,
    lwd = 0.2
  ) +
  #geom_line(data = table_res, aes(x=method,y=mean_RMSE, colour=sensor))+
  # ggtitle("Calibration - PMS - Scenarios comparison") +
  custom_theme +
  theme(axis.text.x = element_text(
    angle = 90,
    hjust = 0.95,
    vjust = 0.2
  )) +
  facet_wrap( ~ sensor_type, ncol = 2, scales = "free_y") +
  xlab("Calibration method") +
  ylab(quickText("Uncertainty USEPA (%)")) +
  ylim(c(0, quantile(res$U, probs = 0.9)))
p_U

## Warning: Removed 13866 rows containing non-finite values (stat_boxplot).

## Warning: Removed 13866 rows containing non-finite values (stat_boxplot).

p_cv <- res %>%
  dplyr::filter(type == "Full") %>%
  dplyr::filter(dataset_type == "cal") %>%
  ggplot(aes(x = method, y = cv, colour = scenario_type)) +
  stat_boxplot(geom = "errorbar") +
  geom_boxplot(
    fatten = 0.8,
    outlier.size = 0.3,
    outlier.alpha = 0.5,
    lwd = 0.2
  ) +
  #geom_line(data = table_res, aes(x=method,y=mean_RMSE, colour=sensor))+
  # ggtitle("Calibration - PMS - Scenarios comparison") +
  custom_theme +
  theme(axis.text.x = element_text(
    angle = 90,
    hjust = 0.95,
    vjust = 0.2
  )) +
  facet_wrap( ~ sensor_type, ncol = 2, scales = "free_y") +
  xlab("Calibration method") +
  ylab(quickText("CV (%)")) +
  ylim(c(0, quantile(res$cv, probs = 0.9)))
p_cv

## Warning: Removed 13563 rows containing non-finite values (stat_boxplot).

## Warning: Removed 13563 rows containing non-finite values (stat_boxplot).

p_ue <- res %>%
  dplyr::filter(type == "Full") %>%
  dplyr::filter(dataset_type == "cal") %>%
  ggplot(aes(x = method, y = expanded_uncertainty, colour = scenario_type)) +
  stat_boxplot(geom = "errorbar") +
  geom_boxplot(
    fatten = 0.8,
    outlier.size = 0.3,
    outlier.alpha = 0.5,
    lwd = 0.2
  ) +
  #geom_line(data = table_res, aes(x=method,y=mean_RMSE, colour=sensor))+
  # ggtitle("Calibration - PMS - Scenarios comparison") +
  custom_theme +
  theme(axis.text.x = element_text(
    angle = 90,
    hjust = 0.95,
    vjust = 0.2
  )) +
  facet_wrap( ~ sensor_type, ncol = 2, scales = "free_y") +
  xlab("Calibration method") +
  ylab(quickText("Expanded uncertainty (%)")) +
  ylim(c(0, quantile(
    res$expanded_uncertainty, probs = 0.9, na.rm = T
  )))
p_ue

## Warning: Removed 11601 rows containing non-finite values (stat_boxplot).

## Warning: Removed 11601 rows containing non-finite values (stat_boxplot).

p_r2 <- res %>%
  dplyr::filter(type == "Full") %>%
  dplyr::filter(dataset_type == "cal") %>%
  ggplot(aes(x = method, y = r2, colour = scenario_type)) +
  stat_boxplot(geom = "errorbar") +
  geom_boxplot(
    fatten = 0.8,
    outlier.size = 0.3,
    outlier.alpha = 0.5,
    lwd = 0.2
  ) +
  #geom_line(data = table_res, aes(x=method,y=mean_RMSE, colour=sensor))+
  # ggtitle("Calibration - PMS - Scenarios comparison") +
  custom_theme +
  theme(axis.text.x = element_text(
    angle = 90,
    hjust = 0.95,
    vjust = 0.2
  )) +
  facet_wrap( ~ sensor_type, ncol = 2, scales = "free_y") +
  xlab("Calibration method") +
  ylab(quickText("r2")) +
  ylim(c(0, quantile(
    res$r2, probs = 0.9, na.rm = T
  )))
p_r2

## Warning: Removed 10373 rows containing non-finite values (stat_boxplot).

## Warning: Removed 10373 rows containing non-finite values (stat_boxplot).

Results in tables

table_res <- res %>%
  dplyr::filter(grepl("PMS", sensor)) %>%
  dplyr::filter(type == "Full") %>%
  dplyr::filter(dataset_type == "cal") %>%
  group_by(scenario_type, sensor, method) %>%
  summarise(
    mean_RMSE = mean(RMSE, na.rm = T),
    quant25 = quantile(RMSE, probs = c(0.25), na.rm = T),
    quant75 = quantile(RMSE, probs = c(0.75), na.rm = T),
    median = median(RMSE, na.rm = T)
  ) %>%
  arrange(desc(mean_RMSE)) #%>%

## `summarise()` has grouped output by 'scenario_type', 'sensor'. You can override using the `.groups` argument.

table_res

table_res %>%
  write.csv(paste0(PLOT_OUTPUT_FOLDER, "table_res_pms.csv"), sep = "\tab")

## Warning in write.csv(., paste0(PLOT_OUTPUT_FOLDER, "table_res_pms.csv"), :
## attempt to set 'sep' ignored

res %>%
  dplyr::filter(grepl("PMS", sensor)) %>%
  dplyr::filter(type == "Full") %>%
  dplyr::filter(dataset_type == "cal") %>%
  group_by(scenario_type, sensor, method) %>%
  rstatix::get_summary_stats()

res_sensor_type <-  res %>%
  dplyr::filter(grepl("PMS", sensor)) %>%
  dplyr::filter(type == "Full") %>%
  dplyr::filter(dataset_type == "cal") %>%
  group_by(scenario_type, method, sensor_type) %>%
  rstatix::get_summary_stats()


table_res_sps <- res %>%
  dplyr::filter(grepl("SPS", sensor)) %>%
  dplyr::filter(type == "Full") %>%
  dplyr::filter(dataset_type == "cal") %>%
  group_by(scenario_type, sensor, method) %>%
  summarise(
    mean_RMSE = mean(RMSE, na.rm = T),
    quant25 = quantile(RMSE, probs = c(0.25), na.rm = T),
    quant75 = quantile(RMSE, probs = c(0.75), na.rm = T),
    median = median(RMSE, na.rm = T)
  ) %>%
  arrange(desc(mean_RMSE))

## `summarise()` has grouped output by 'scenario_type', 'sensor'. You can override using the `.groups` argument.

table_res_sps

table_res_sps %>%
  write.csv(paste0(PLOT_OUTPUT_FOLDER, "table_res_sps.csv"), sep = "\tab")

## Warning in write.csv(., paste0(PLOT_OUTPUT_FOLDER, "table_res_sps.csv"), :
## attempt to set 'sep' ignored

Calibration scenarios comparison

FMJ Bulot (florentin.bulot@centraliens.net)

Last generated: 25 October, 2022, 11:43