CODEWITHSUNDEEP
rif-statementerror-handlingshinytry-catch
Mary
Mary

Reputation: 127

Warning handling with if/else blocks in Shiny

I am developing a shiny app. which selects nominal (factors) and ordinal(numeric) variables for a given dataset. Then it transform nominal variable(s) (male,female) to dummy variable(s). And eventually merges ordinal variables with dummy variables. The app works fine if I select both nominal and ordinal variables. But if I only want to choose nominal OR ordinal variables,I get warnings: "No character or factor columns found. Please use select_columns to choose columns." Here are my server.r and ui.R

library(fastDummies)
library(data.table)
data=data.frame(A=c(5,2,4),B=c('male','male','female'),C=c(1,3,5))
shinyServer(
function(input, output){
#Select Explanatory Nominal Variables 
output$ColumnSelector_dummy <- renderUI({
  selectInput("SelectedDummy","Select Nominal Variables (values: male,female)", 
              choices = as.list(names(data)),multiple=TRUE, selected = NULL)
})
#
df_subset_dummy <- reactive({
  a <- subset(data, select = input$SelectedDummy)
  return(a) 
})
#Convert Nominal variables to Dummy variables 
df_subset_dummy_tranformed <- reactive({
  df1 <- lapply( df_subset_dummy(), factor)
  df2 <- fastDummies::dummy_cols(df1)
  drops <- names(df1)
  df3 <- df2[, !(names(df2) %in% drops)]
  return(df3)
})
#Select Explanatory Ordinal Variables
output$ColumnSelector_ordinal<- renderUI({
  selectInput("SelectedOrdinal","Select Ordinal Variables (values: 1,2,3,4,5,6)", 
              choices = as.list(names(data)), multiple=TRUE,selected = NULL )
}) 
 df_subset_ordinal <- reactive({
  a <- subset(data, select = input$SelectedOrdinal)
  return(a) 
})  
 #Join Ordinal and Nominal dataframes   
df_nominal_ordinal_bind <- reactive({  
  df <- cbind(df_subset_dummy_tranformed(),df_subset_ordinal()) 
  return(df)
})
output$table_ordinal_nominal <- renderTable(head(df_nominal_ordinal_bind()))  
})

shinyUI(
 fluidPage(
   tabsetPanel(
     tabPanel("Data", fluid = TRUE,
            sidebarLayout(
               sidebarPanel(
               uiOutput("ColumnSelector_dummy"),  
               uiOutput("ColumnSelector_ordinal")
             ),
             mainPanel(
               tabsetPanel(
                 tabPanel('Subsets',
                          tableOutput('table_ordinal_nominal')
               ) )) ) ))))

question: How can I use tryCatch and if/else blockes, so that if df_subset_dummy_tranformed() OR df_subset_ordinal() does not exist, I still get result for df_nominal_ordinal_bind and it also can be shown as a table in output$table_ordinal_nominal. Any help would be appreciated.

Upvotes: 0

Views: 56

Answers (1)

RolandASc
RolandASc

Reputation: 3923

Without changing your code / approach too much, I guess you could start by adding a check on df_subset_dummy():

df_subset_dummy_tranformed <- reactive({
  res <- df_subset_dummy()
  if (length(res) == 0) return(res)
  df1 <- lapply(res, factor)
  df2 <- fastDummies::dummy_cols(df1)
  drops <- names(df1)
  df3 <- df2[, !(names(df2) %in% drops)]
  return(df3)
})

At least this should address the warning you mention.

Upvotes: 1

Related Questions