How to remove NA form the legend?

[MathieuVandamme]

Hi all

I am making a plot that represents the total amount of samples of each machine over 24 hours. You can choose your systems via a checkbos group.

I want to give different colors to the different machines. But a some hours, there are no samples don an thus 0 in the input dataframe for the ggplot. This 0 come in the legend, but I want to get rid of it. I already changed it to NA, it remains in the plot. I made an extra dataframe for the legend, and removed the NA, but then I shows the one selected machine twice.

Here is my code for the plot (especially the last chunk is used for the plot):

 #Calculate the amount of Result-samples each hour:
      hours_set <- hms(df1_filtered$ResultTime)
      df2 <- as.data.frame(hours_set$hour)
      df2["system"] <- df1_filtered$InstrumentName
      df_aggr_Result <- aggregate(df2, by=list(df2$`hours_set$hour`, df2$system), FUN = length)
      #Renaming the columns:
      names(df_aggr_Result)[names(df_aggr_Result) == "Group.1"] <- "hour"
      names(df_aggr_Result)[names(df_aggr_Result) == "hours_set$hour"] <- "amount of samples"
      names(df_aggr_Result)[names(df_aggr_Result) == "Group.2"] <- "system"
      
      update_busy_bar(66.66)  
      #Make sure that there are always 24 hours in a dataset, even if there are no counts for that hour:
      all_h <- tibble(hour = 0:23)
      df_plot = merge(x=all_h,y=df_aggr_Result,by="hour",all=TRUE)
      df_plot[is.na(df_plot)] <- 0
      df_plot$system[df_plot$system == 0] <- NA
      #Create the labels in total amount and percentage:
      df_plot <-
        mutate(df_plot, p = df_plot$`amount of samples`/ sum(df_plot$`amount of samples`),
               p = scales::label_percent(accuracy = 0.01)(p),
               lab = paste(df_plot$`amount of samples`, p, sep = "\n"))
      #Remove the NA for the legend:
      df_plot_legend <- df_plot[!is.na(df_plot$system),]
      
      update_busy_bar(99.99)
      update_busy_bar(100)
      ggplot(df_plot, aes(x = df_plot$hour, y = df_plot$`amount of samples`, fill = df_plot$system)) +
        geom_bar(stat = "identity", position="stack") +
        theme(axis.text.x = element_text(face = "bold", color = "#993333", size = 15),
              axis.text.y = element_text(face = "bold", color = "#993333", size = 15),
              axis.line = element_line(color = "#993333", size = 1)) +
        scale_x_continuous(breaks=seq(0,23,1)) +
        scale_y_continuous(limits = c(0,max(df_plot$`amount of samples`)+(0.1*max(df_plot$`amount of samples`)))) + #Add another 10% of the max to the y-scale to improve visual of the labels.
        xlab("Hours in a day") + ylab("Amount of samples") +
        guides(fill = guide_legend(title = "System:")) +
        scale_fill_hue(labels = df_plot_legend$system)
      #geom_text(
          #label= if_else(df_plot$`amount of samples` > 0, df_plot$lab, ""), 
          #vjust = -0.25,
          #nudge_y = 0.1, 
          #check_overlap = F
        #)
    })    

How can I remove it? I already look on the internet, but it don't seems to work out for me.

Thanks in advance!

[nirgrahamuk]

filter your data.frame so it doesnt include entries with amount of samples of zero ?
This is a tentative suggestion, as you have not provided a reprex

[Flm]

Take a look: How to remove NA (data with missing values) in geom_col? - #2 by Flm

The easiest way is to change the first line ggplot(df_plot, aes(...)) into

df_plot %>%
  drop_na() %>%
  ggplot(aes(x = df_plot$hour, y = df_plot$`amount of samples`, fill = df_plot$system)) +

[MathieuVandamme]

@MathieuVandamme

ggplot(df_plot, aes(x = df_plot$hour, y = df_plotamount of samples, fill = df_plot$system)) +

Both don't seems to work. The plot is based on this table and as long it contains "NA", NA will be in the legend. Is there a work-around?

Thank you!

[Flm]

Can you provide part of your df using this command: dput(head(YOURDF, 20))?

---

EDIT:
oh, I think this should solve the problem:

df_plot %>%
  drop_na() %>%
  ggplot(aes(x = hour, y = `amount of samples`, fill = system)) +

Because now it is in a pipeline df_plot$ have to be removed. Remove it also in scale_y_continuous, scale_fill_hue etc...

[MathieuVandamme]

Hi all

That did the trick. It worked! Thank you!

This is my final code:

      #Calculate the amount of Result-samples each hour:
      hours_set <- hms(df1_filtered$ResultTime)
      df2 <- as.data.frame(hours_set$hour)
      df2["system"] <- df1_filtered$InstrumentName
      df_aggr_Result <- aggregate(df2, by=list(df2$`hours_set$hour`, df2$system), FUN = length)
      #Renaming the columns:
      names(df_aggr_Result)[names(df_aggr_Result) == "Group.1"] <- "hour"
      names(df_aggr_Result)[names(df_aggr_Result) == "hours_set$hour"] <- "amount of samples"
      names(df_aggr_Result)[names(df_aggr_Result) == "Group.2"] <- "system"
      
      update_busy_bar(66.66)  
      #Make sure that there are always 24 hours in a dataset, even if there are no counts for that hour:
      all_h <- tibble(hour = 0:23)
      df_plot = merge(x=all_h,y=df_aggr_Result,by="hour",all=TRUE)
      df_plot[is.na(df_plot)] <- 0
      df_plot$system[df_plot$system == 0] <- NA
      #Create the labels in total amount and percentage:
      df_plot <-
        mutate(df_plot, p = df_plot$`amount of samples`/ sum(df_plot$`amount of samples`),
               p = scales::label_percent(accuracy = 0.01)(p))
      
      
      output$labels_system1 <- renderTable({
        df_plot_table <- data.frame(select(df_plot, -system.1))
        #df_plot_table <-setNames(df_plot_table, rep(" ", length(df_plot_trans)))
        colnames(df_plot_table) <- c("Hour","System","Amount of samples","Percentage")
        df_plot_table
      }, rownames = FALSE)
      
      update_busy_bar(99.99)
      update_busy_bar(100)
      df_plot %>%
        drop_na() %>%
        ggplot(aes(x = hour, y = `amount of samples`, fill = system)) +
        geom_bar(stat = "identity", position="stack") +
        theme(axis.text.x = element_text(face = "bold", color = "#993333", size = 15),
              axis.text.y = element_text(face = "bold", color = "#993333", size = 15),
              axis.line = element_line(color = "#993333", size = 1)) +
        scale_x_continuous(breaks=seq(0,23,1)) +
        scale_y_continuous(limits = c(0,max(df_plot$`amount of samples`)+(0.1*max(df_plot$`amount of samples`)))) + #Add another 10% of the max to the y-scale to improve visual of the labels.
        xlab("Hours in a day") + ylab("Amount of samples") +
        guides(fill = guide_legend(title = "System:"))
    })