update: explo_ratios_trend

inst/main_bdown/exploration_ratios_trend.Rmd

+# Explore data trend
+
+## Methodology
+
+on these plots we continue to explore the significant variables. we seek to simplify the reading of the boxplots by creating for each plot :
+
+- A line representing the average increase in the value of patients in the non_resistant group. 
+The points at visit S and visit 2 are respectively the mean of the values at visit 1 and visit 2 (only non-resistant patients with a positive ratio). 
+
+- A line showing the average decrease in the value of patients in the non_resistant group. 
+The points at Visit S and Visit 2 are respectively the mean values at Visit 1 and Visit 2 (only non-resistant patients with a negative ratio).
+
+- A line showing the average increase in the value of patients in the primary_resistant group. 
+The points at Visit S and Visit 2 are respectively the mean of the values at Visit 1 and Visit 2 (only primary_resistant patients with a positive ratio).
+
+- A line showing the average decrease in the value of patients in the primary_resistant group. 
+The points at Visit S and Visit 2 are respectively the average of the values at Visit 1 and Visit 2 (only primary_resistant patients with a negative ratio).
+
+## Connected boxplots
+
+### Very significant (t-test, $p < 0.01$):
+
+```{r connected_boxplots_trends very signif, cache=FALSE, echo=FALSE, message=FALSE, warning=FALSE}
+#display complete subjid in my init work dataframe
+df_boxplots = .df_bm_original %>%
+  select(SUBJID, any_of(dictionary_ratios$CODE_from)) %>%
+  rename_all(~ c("SUBJID", dictionary_ratios %>% pull(CODE) %>% as.character())) %>%
+  mutate(VISIT = from) %>%
+  rbind(
+    .df_bm_original %>%
+      select(SUBJID, any_of(dictionary_ratios$CODE_to)) %>%
+      rename_all(~ c("SUBJID", dictionary_ratios %>% pull(CODE) %>% as.character())) %>%
+      mutate(VISIT = to)
+  ) 
+
+# creation work dataframe very significant
+df_boxplots_trend_very_signif <- df_boxplots %>%
+                     select(any_of(c('SUBJID',dictionary_very_signif$CODE,'VISIT')))%>%
+                     rename_with(~paste0(., "_values"), c(-VISIT, -SUBJID))
+
+
+# keep ratio of very significant variables
+df_ratio_trend_very_signif <- ratios$df %>%
+  select(any_of(c('SUBJID', dictionary_very_signif$CODE))) 
+
+
+# merge SUIBJID, VISIT, VS_to_V2, Ratios
+.df_resistance = subset(.df_outcome, select=c(SUBJID, PRIM_RES))
+df_boxplots_trend_very_signif2= merge(df_boxplots_trend_very_signif, .df_resistance, by = "SUBJID")
+df_boxplots_trend_very_signif3= merge(df_boxplots_trend_very_signif2, df_ratio_trend_very_signif, by = "SUBJID")
+```
+
+
+```{r create data very signif}  
+
+variables_values=c(colnames(df_boxplots_trend_very_signif2))
+variables_values=variables_values[variables_values!= "SUBJID" & variables_values!= "VISIT" & variables_values!= "PRIM_RES"]
+variables_values = lapply(variables_values, sym)
+
+variables_ratio=c(colnames(df_ratio_trend_very_signif))
+variables_ratio=variables_ratio[variables_ratio!= "SUBJID" & variables_ratio!= "VISIT" & variables_ratio!= "PRIM_RES"]
+variables_ratio = lapply(variables_ratio, sym)
+```
+
+
+```{r Prepare trend data median, results = "asis", cache=FALSE, echo=FALSE, message=FALSE, warning=FALSE} 
+
+compute_median_data = function(df, x_var, y_var) { 
+  median_data = df %>%
+    group_by(PRIM_RES, {{ x_var }} > 0, VISIT) %>%
+    summarise(median = median({{ y_var }}))
+    median_data$PRIM_RES <- factor(median_data$PRIM_RES, levels = c(0, 1), labels = c("non resistant", "primary resistant"))
+  return(median_data)
+}
+
+ponderation = function(df, x_var) {
+count_data <- df %>%
+  group_by(PRIM_RES, {{ x_var }} > 0, VISIT) %>%
+  summarise(patient_count = n())
+return(count_data)
+}
+
+plot_median_data2 = function(tab_median,tab_ponderation,ratio_label) {
+
+  label_legend = ratio_label
+  label_legend = dictionary_ratios$LABEL[dictionary_ratios$CODE == ratio_label]
+  
+  tab_median = na.omit(tab_median)
+  tab_ponderation = na.omit(tab_ponderation)
+  tab_median$index = c(rep(1:(nrow(tab_median)/2), each = 2))
+
+  median_fig = ggplot(tab_median, aes(x = factor(VISIT, levels = c("VS", "V2")), y = median, color = PRIM_RES, group = tab_median[[2]])) +
+    geom_point()+
+    geom_line(aes(group = tab_median$index ), size = (tab_ponderation$patient_count/4))+
+    geom_text(aes(label = paste("nb:", tab_ponderation$patient_count), y = median), hjust = -0.2, vjust = 0) +
+    
+      labs(title = label_legend,
+         x = "Visite",
+         color = "Resistance") +
+    theme_minimal()
+  return(median_fig)
+ 
+}
+
+```
+
+
+
+```{r Prepare trend data mean, results = "asis", cache=FALSE, echo=FALSE, message=FALSE, warning=FALSE} 
+
+compute_mean_data = function(df, x_var, y_var) { 
+  mean_data = df %>%
+    group_by(PRIM_RES, {{ x_var }} > 0, VISIT) %>%
+    summarise(mean = mean({{ y_var }}))
+    mean_data$PRIM_RES <- factor(mean_data$PRIM_RES, levels = c(0, 1), labels = c("non resistant", "primary resistant"))
+  return(mean_data)
+}
+
+ponderation = function(df, x_var) {
+count_data <- df %>%
+  group_by(PRIM_RES, {{ x_var }} > 0, VISIT) %>%
+  summarise(patient_count = n())
+return(count_data)
+}
+
+plot_mean_data2 = function(tab_mean,tab_ponderation,ratio_label) {
+
+  label_legend = ratio_label
+  label_legend = dictionary_ratios$LABEL[dictionary_ratios$CODE == ratio_label]
+  
+  tab_mean = na.omit(tab_mean)
+  tab_ponderation = na.omit(tab_ponderation)
+  tab_mean$index = c(rep(1:(nrow(tab_mean)/2), each = 2))
+
+  mean_fig = ggplot(tab_mean, aes(x = factor(VISIT, levels = c("VS", "V2")), y = mean, color = PRIM_RES, group = tab_mean[[2]])) +
+    geom_point()+
+    geom_line(aes(group = tab_mean$index ), size = (tab_ponderation$patient_count/4))+
+    geom_text(aes(label = paste("nb:", tab_ponderation$patient_count), y = mean), hjust = -0.2, vjust = 0) +
+    
+      labs(title = label_legend,
+         x = "Visite",
+         color = "Resistance") +
+    theme_minimal()
+  return(mean_fig)
+ 
+}
+
+```
+
+
+#### Median trend
+
+```{r plot trend data median very signif, results = "asis", cache=FALSE, echo=FALSE, message=FALSE, warning=FALSE} 
+
+tab_median = pmap(list(
+    rep(list(df_boxplots_trend_very_signif3), length(variables_values)), 
+       variables_ratio, 
+       variables_values), 
+  compute_median_data) 
+
+tab_ponderation = pmap(list(
+    rep(list(df_boxplots_trend_very_signif3), length(variables_values)), 
+    variables_ratio),
+  ponderation)
+
+plot_median = pmap(list(
+  tab_median, tab_ponderation,variables_ratio), plot_median_data2)
+ 
+#plot_median
+
+
+
+compoEDA::eda_multiplot(
+  plot_median,
+  ncol      = 4,
+  text_size = 4
+ )
+
+```
+
+
+#### Mean trend
+
+```{r plot trend data mean very signif, results = "asis", cache=FALSE, echo=FALSE, message=FALSE, warning=FALSE} 
+
+tab_mean = pmap(list(
+    rep(list(df_boxplots_trend_very_signif3), length(variables_values)), 
+       variables_ratio, 
+       variables_values), 
+  compute_mean_data) 
+
+tab_ponderation = pmap(list(
+    rep(list(df_boxplots_trend_very_signif3), length(variables_values)), 
+    variables_ratio),
+  ponderation)
+
+plot_mean = pmap(list(
+  tab_mean, tab_ponderation,variables_ratio), plot_mean_data2)
+ 
+#plot_median
+
+
+
+compoEDA::eda_multiplot(
+   plot_mean,
+   ncol      = 4,
+   text_size = 4
+  )
+
+```
+
+### Significant (t-test, $0.01 \leq p < 0.05$):
+
+```{r connected_boxplots_trends, cache=FALSE, echo=FALSE, message=FALSE, warning=FALSE}
+
+# creation work dataframe significant
+df_boxplots_trend <- df_boxplots %>%
+                     select(any_of(c('SUBJID',dictionary_signif$CODE,'VISIT')))%>%
+                     rename_with(~paste0(., "_values"), c(-VISIT, -SUBJID))
+
+
+# keep ratio of significant variables
+df_ratio_trend <- ratios$df %>%
+  select(any_of(c('SUBJID', dictionary_signif$CODE))) 
+
+
+
+# merge SUIBJID, VISIT, VS_to_V2, Ratios
+.df_resistance = subset(.df_outcome, select=c(SUBJID, PRIM_RES))
+df_boxplots_trend2= merge(df_boxplots_trend, .df_resistance, by = "SUBJID")
+df_boxplots_trend3= merge(df_boxplots_trend2, df_ratio_trend, by = "SUBJID")
+```
+
+
+```{r create data}  
+
+variables_values=c(colnames(df_boxplots_trend2))
+variables_values=variables_values[variables_values!= "SUBJID" & variables_values!= "VISIT" & variables_values!= "PRIM_RES"]
+variables_values = lapply(variables_values, sym)
+
+variables_ratio=c(colnames(df_ratio_trend))
+variables_ratio=variables_ratio[variables_ratio!= "SUBJID" & variables_ratio!= "VISIT" & variables_ratio!= "PRIM_RES"]
+variables_ratio =lapply(variables_ratio, sym)
+```
+
+#### Median trend
+
+```{r plot trend data median, results = "asis", cache=FALSE, echo=FALSE, message=FALSE, warning=FALSE} 
+
+tab_median = pmap(list(
+    rep(list(df_boxplots_trend3), length(variables_values)), 
+       variables_ratio, 
+       variables_values), 
+  compute_median_data) 
+
+tab_ponderation = pmap(list(
+    rep(list(df_boxplots_trend3), length(variables_values)), 
+    variables_ratio),
+  ponderation)
+
+plot_median = pmap(list(
+  tab_median, tab_ponderation,variables_ratio), plot_median_data2)
+ 
+#plot_median
+
+
+
+compoEDA::eda_multiplot(
+   plot_median,
+   ncol      = 4,
+   text_size = 4
+  )
+
+```
+
+
+#### Mean trend
+
+
+```{r plot trend data mean, results = "asis", cache=FALSE, echo=FALSE, message=FALSE, warning=FALSE} 
+
+tab_mean = pmap(list(
+    rep(list(df_boxplots_trend3), length(variables_values)), 
+       variables_ratio, 
+       variables_values), 
+  compute_mean_data) 
+
+tab_ponderation = pmap(list(
+    rep(list(df_boxplots_trend3), length(variables_values)), 
+    variables_ratio),
+  ponderation)
+
+plot_mean = pmap(list(
+  tab_mean, tab_ponderation,variables_ratio), plot_mean_data2)
+ 
+#plot_median
+
+
+
+compoEDA::eda_multiplot(
+   plot_mean,
+   ncol      = 4,
+   text_size = 4
+  )
+
+```
\ No newline at end of file