quick save before swtich branch

lib/2_loq_extractions.R

 ## extract LOQs for the data
 extract_loqs <- function(df, list_pfas){
   data_loq <- df %>%
-    filter(value_below_loq == TRUE) %>%
     group_by(substance) %>%
     summarise(min_loq = min(loq),
               max_loq = max(loq), .groups = "drop") %>%

src/3_MFA.R

-## multi-
+## Factor Analysis on Mixed Data (FAMD)
 library(ProjectTemplate)
 load.project()
 
 library(FactoMineR)
-library(factoextra)
-
-# ------------ PCA --------------
-## standardization
-data_corr <- data_ros %>%
-  left_join(info_pfas, by = c("substance", "cas_number")) %>%
-  group_by(type, substance) %>%
-  mutate(value_scaled = scale(value))
-  
-numeric_data_scaled <- data_corr$value_scaled
-
-## perform pca
-pca_result <- PCA(numeric_data_scaled, graph = FALSE)
+library(Factoshiny)
+
+# -------- restructure the data for FAMD --------
+data_famd_raw <- data_ros %>%
+  filter(datasource == "EU-PROMISCES") %>%
+  select(code, matrix, type, substance, value) %>%
+  group_by(code, matrix, type, substance) %>%
+  filter(row_number() == 1) %>%
+  ungroup() %>%
+  pivot_wider(names_from = substance, values_from = value)
+
+colSums(is.na(data_famd_raw))
+# data_famd_raw %>% filter(if_any(4:35, is.na))
+
+# fill NAs with half_loq
+data_loq <- data_ros %>%
+    filter(datasource == "EU-PROMISCES") %>%
+    filter(value_below_loq == TRUE) %>%
+    group_by(substance, type) %>%
+    summarise(loq = mean(loq), .groups = "drop") 
+
+data_famd_long <- data_famd_raw %>%
+  pivot_longer(cols = 4:35, names_to = "substance", values_to = "value") %>%
+  left_join(data_loq, by = c("substance", "type")) %>%
+  mutate(value = ifelse(is.na(value), loq / 2, value)) %>%
+  select(-loq)
+
+data_famd <- data_famd_long %>%
+  pivot_wider(names_from = substance, values_from = value) %>%
+  select(-code)
+
+Factoshiny(data_famd)
+
+# -------- perform MFA ------
+res.FAMD<-FAMD(data_famd,graph=FALSE)
+plot.FAMD(res.FAMD,invisible=c('ind.sup'),title="Graph of individuals and categories")
+plot.FAMD(res.FAMD,axes=c(1,2),choix='var',title="Graph of the variables")
+plot.FAMD(res.FAMD, choix='quanti',title="Correlation circle")
+
+
+
+summary(res.FAMD)
+

src/4_boxplots.R

@@ -2,4 +2,18 @@
 library(ProjectTemplate)
 load.project()
 
+# ------- SW and GW ------
+## define list of type and pfas to present
+type_swgw <- level_type[c(2,3,4,5,6)]
+pfas_swgw <- c("PFBA", "PFPeA", "PFHxA", "PFHpA", "PFOA",
+               "PFBS", "PFHxS", "PFOS", "GenX", "ADONA")
+## prepare dataset
+data_swgw <- data_ros %>%
+  filter(type %in% type_swgw, substance %in% pfas_swgw) %>%
+  droplevels()
+loq_swgw <- extract_loqs(data_swgw, pfas_swgw)
 
+test <- data_swgw %>% arrange(desc(loq))
+
+test2 <- data_all %>% filter(datasource == "AT-GZUV-GW") 
+table(test2$loq)