I have recently been exploring shiny and shinydashboard using the gapminder dataset and have encountered my latest issue trying to output the summary of a reactive model created using a response variable and predictor variable based on user selections. The aim is to have both EDA plots and summaries for 3 models output concurrently (base, log and inverse) which is why var_EDA_plots() is a little messy. The error that is being returned is:
# Error: argument 1 (type 'list') cannot be handled by 'cat'
The code relevant to the simple linear regression modelling is:
UI
# LM Page Content
tabItem(tabName = "lm",
h2("Simple Linear Regression"),
h3(wellPanel(fluidRow(
column(2,"Response Variable:"),
column(2, selectInput("response", "", colnames(data)[4:6],
multiple = FALSE, selected = "lifeExp")),
column(2, "Predictor Variable: "),
column(2, selectInput("predictor", "", colnames(data)[4:6],
multiple = FALSE, selected = "gdpPercap"))))
),
tabsetPanel(type = "tabs",
tabPanel("Plot",
splitLayout(
plotOutput("EDAplotBase"),
plotOutput("EDAplotLog"),
plotOutput("EDAplotInv"))),
tabPanel("Summary", verbatimTextOutput("base.lm")),
tabPanel("Table", tableOutput("EDAtable"))
))
SERVER
# LINEAR MODELLING ----
# Format Predictor and Response col names
var_EDA_plots <- reactive({
str_col_predictor <- as.character(input$predictor)
str_col_response <- as.character(input$response)
transformed_data <- data %>% select(str_col_response, str_col_predictor, continent, country, year)
transformed_data$predictor_base <- unlist(transformed_data[2])
transformed_data$predictor_log <- unlist(log(transformed_data[2]))
transformed_data$predictor_inv <- unlist(1/transformed_data[2])
transformed_data$response <- unlist(transformed_data[1])
transformed_data %>% select(continent, country, year, response, predictor_base, predictor_log, predictor_inv)
})
# Build linear models reactively
var_lm_base <- reactive({
fml <- as.formula("response ~ predictor_base")
lm(fml, data = var_EDA_plots())
})
# Base LM Model
output$base.lm <- renderText({
summary(var_lm_base())
})
output$EDAtable <- renderTable({var_EDA_plots()})
Below I'll paste the entire code for the shiny dashboard if it is easier to debug and test in it's entirety (linear modelling section is at the bottom of server):
library(shiny)
library(tidyverse)
library(shinydashboard)
library(gapminder)
# LOAD DATA ----
data <- gapminder %>% as_tibble() %>% arrange(country, year)
# LINEAR MODELLING ----
set.seed(117)
train <- data %>% slice_sample(prop = 0.8)
test <- data %>% slice_sample(prop = 0.2)
# UI ----
ui <- dashboardPage(
dashboardHeader(title = "Gapminder Dashboard"),
dashboardSidebar(
sidebarMenu(id = "tabs",
menuItem("Dashboard", tabName = "dashboard", icon = icon("dashboard"),
menuSubItem("Life Expectancy", tabName = "life"),
menuSubItem("GDP Per Capita", tabName = "gdp")),
menuItem("Linear Modelling", icon = icon("th"), tabName = "lm", badgeLabel = "new", badgeColor = "green"),
fluidPage(
selectInput("dateStart", "Start date:", distinct(data, year),
multiple = FALSE, selected = 1952),
selectInput("dateEnd", "End date:", distinct(data, year),
multiple = FALSE, selected = 2007),
selectInput("country1", "Select primary country:", distinct(data, country),
multiple = FALSE, selected = "Australia"),
selectInput("country2", "Select secondary country:", distinct(data, country),
multiple = FALSE, selected = "Greece")
))),
dashboardBody(
tabItems(
# Life Expectancy Page Content
tabItem(tabName = "life",
# Top 5 KPIs
splitLayout(
valueBoxOutput("kpi.top5.life1", width = NULL),
valueBoxOutput("kpi.top5.life2", width = NULL),
valueBoxOutput("kpi.top5.life3", width = NULL),
valueBoxOutput("kpi.top5.life4", width = NULL),
valueBoxOutput("kpi.top5.life5", width = NULL)
),
# Life Expectancy Histogram
fluidPage(
plotOutput("histLifeExp")
),
# Life Expectancy & Population Plots
splitLayout(
plotOutput("lineplotLife"),
plotOutput("lineplotPopn")
),
# Bottom 5 KPIs
splitLayout(
valueBoxOutput("kpi.btm5.life1", width = NULL),
valueBoxOutput("kpi.btm5.life2", width = NULL),
valueBoxOutput("kpi.btm5.life3", width = NULL),
valueBoxOutput("kpi.btm5.life4", width = NULL),
valueBoxOutput("kpi.btm5.life5", width = NULL)
)),
# GDP Page Content
tabItem(tabName = "gdp",
# Top 5 KPIs
splitLayout(
valueBoxOutput("kpi.top5.gdp1", width = NULL),
valueBoxOutput("kpi.top5.gdp2", width = NULL),
valueBoxOutput("kpi.top5.gdp3", width = NULL),
valueBoxOutput("kpi.top5.gdp4", width = NULL),
valueBoxOutput("kpi.top5.gdp5", width = NULL)
),
# GDP Histogram
fluidPage(
plotOutput("histGDP")
),
# GDP & Population Plots
splitLayout(
plotOutput("lineplotgdp"),
plotOutput("lineplotPopn2")
),
# Bottom 5 KPIs
splitLayout(
valueBoxOutput("kpi.btm5.gdp1", width = NULL),
valueBoxOutput("kpi.btm5.gdp2", width = NULL),
valueBoxOutput("kpi.btm5.gdp3", width = NULL),
valueBoxOutput("kpi.btm5.gdp4", width = NULL),
valueBoxOutput("kpi.btm5.gdp5", width = NULL)
)),
# LM Page Content
tabItem(tabName = "lm",
h2("Simple Linear Regression"),
h3(wellPanel(fluidRow(
column(2,"Response Variable:"),
column(2, selectInput("response", "", colnames(data)[4:6],
multiple = FALSE, selected = "lifeExp")),
column(2, "Predictor Variable: "),
column(2, selectInput("predictor", "", colnames(data)[4:6],
multiple = FALSE, selected = "gdpPercap"))))
),
tabsetPanel(type = "tabs",
tabPanel("Plot",
splitLayout(
plotOutput("EDAplotBase"),
plotOutput("EDAplotLog"),
plotOutput("EDAplotInv"))),
tabPanel("Summary", verbatimTextOutput("base.lm")),
tabPanel("Table", tableOutput("EDAtable"))
))
)
)
)
# SERVER ----
server <- function(input, output) {
# REACTIVE DATA FILTERING ----
# Top 5 Life Exp KPIs - date filtering
var_maxDate_kpi_top5_life <- reactive({
val <- as.integer(input$dateEnd)
data %>% filter(year == val) %>% slice_max(n = 5, order_by = lifeExp)
})
# Bottom 5 Life Exp KPIs - date filtering
var_maxDate_kpi_btm5_life <- reactive({
val <- as.integer(input$dateEnd)
data %>% filter(year == val) %>% slice_min(n = 5, order_by = lifeExp)
})
# Top 5 GDP KPIs - date filtering
var_maxDate_kpi_top5_gdp <- reactive({
val <- as.integer(input$dateEnd)
data %>% filter(year == val) %>% slice_max(n = 5, order_by = gdpPercap)
})
# Bottom 5 GDP KPIs - date filtering
var_maxDate_kpi_btm5_gdp <- reactive({
val <- as.integer(input$dateEnd)
data %>% filter(year == val) %>% slice_min(n = 5, order_by = gdpPercap)
})
# General Life Expectancy Reactive Filtering
var_date_and_country <- reactive({
startVal <- as.integer(input$dateStart)
endVal <- as.integer(input$dateEnd)
country1 <- as.character(input$country1)
country2 <- as.character(input$country2)
data %>% filter(year >= startVal & year <= endVal & country %in% c(country1, country2))
})
# LIFE EXPECTANCY ----
# Value Boxes - Top 5 KPIs | Life Expectancy
output$kpi.top5.life1 <- renderValueBox({
valueBox(paste(round(var_maxDate_kpi_top5_life()$lifeExp[1],1), "years"),
paste(var_maxDate_kpi_top5_life()$country[1], " (", input$dateEnd, ")", sep = ""), icon = icon("heart"), color = "green")
})
output$kpi.top5.life2 <- renderValueBox({
valueBox(paste(round(var_maxDate_kpi_top5_life()$lifeExp[2],1), "years"),
paste(var_maxDate_kpi_top5_life()$country[2], " (", input$dateEnd, ")", sep = ""), icon = icon("heart"), color = "green")
})
output$kpi.top5.life3 <- renderValueBox({
valueBox(paste(round(var_maxDate_kpi_top5_life()$lifeExp[3],1), "years"),
paste(var_maxDate_kpi_top5_life()$country[3], " (", input$dateEnd, ")", sep = ""), icon = icon("heart"), color = "green")
})
output$kpi.top5.life4 <- renderValueBox({
valueBox(paste(round(var_maxDate_kpi_top5_life()$lifeExp[4],1), "years"),
paste(var_maxDate_kpi_top5_life()$country[4], " (", input$dateEnd, ")", sep = ""), icon = icon("heart"), color = "green")
})
output$kpi.top5.life5 <- renderValueBox({
valueBox(paste(round(var_maxDate_kpi_top5_life()$lifeExp[5],1), "years"),
paste(var_maxDate_kpi_top5_life()$country[5], " (", input$dateEnd, ")", sep = ""), icon = icon("heart"), color = "green")
})
# GGPLOT - Life Expectancy
output$lineplotLife <- renderPlot({
ggplot(var_date_and_country(), aes(x = year, y = lifeExp, color = country)) +
geom_line(lwd = 1.5) +
theme_grey() +
labs(x = "Year",
y = "Life Expectancy",
title = paste("Life Expectancy Trend over time (", input$country1, " v ", input$country2, ")", sep = ""))
})
output$lineplotPopn <- renderPlot({
ggplot(var_date_and_country(), aes(x = year, y = (pop/10^6), color = country)) +
geom_line(lwd = 1.5) +
theme_grey() +
labs(x = "Year",
y = "Population (Millions)",
title = paste("Country Population in millions over time (", input$country1, " v ", input$country2, ")", sep = ""))
})
output$histLifeExp <- renderPlot({
ggplot(data, aes(x = lifeExp, color = continent, fill = continent)) +
geom_histogram() +
theme_grey() +
labs(x = "Life Expectancy",
y = "Frequency",
title = "Life Expectancy distribution by Continent")
})
# Value Boxes - Bottom 5 KPIs | Life Expectancy
output$kpi.btm5.life1 <- renderValueBox({
valueBox(paste(round(var_maxDate_kpi_btm5_life()$lifeExp[1],1), "years"),
paste(var_maxDate_kpi_btm5_life()$country[1], " (", input$dateEnd, ")", sep = ""), icon = icon("heart"), color = "red")
})
output$kpi.btm5.life2 <- renderValueBox({
valueBox(paste(round(var_maxDate_kpi_btm5_life()$lifeExp[2],1), "years"),
paste(var_maxDate_kpi_btm5_life()$country[2], " (", input$dateEnd, ")", sep = ""), icon = icon("heart"), color = "red")
})
output$kpi.btm5.life3 <- renderValueBox({
valueBox(paste(round(var_maxDate_kpi_btm5_life()$lifeExp[3],1), "years"),
paste(var_maxDate_kpi_btm5_life()$country[3], " (", input$dateEnd, ")", sep = ""), icon = icon("heart"), color = "red")
})
output$kpi.btm5.life4 <- renderValueBox({
valueBox(paste(round(var_maxDate_kpi_btm5_life()$lifeExp[4],1), "years"),
paste(var_maxDate_kpi_btm5_life()$country[4], " (", input$dateEnd, ")", sep = ""), icon = icon("heart"), color = "red")
})
output$kpi.btm5.life5 <- renderValueBox({
valueBox(paste(round(var_maxDate_kpi_btm5_life()$lifeExp[5],1), "years"),
paste(var_maxDate_kpi_btm5_life()$country[5], " (", input$dateEnd, ")", sep = ""), icon = icon("heart"), color = "red")
})
# GDP PER CAPITA ----
# Value Boxes - Top 5 KPIs | GDP Per Capita
output$kpi.top5.gdp1 <- renderValueBox({
valueBox(paste("$", round(var_maxDate_kpi_top5_gdp()$gdpPercap[1]/1000,1), "k"),
paste(var_maxDate_kpi_top5_gdp()$country[1], " (", input$dateEnd, ")", sep = ""), icon = icon("dollar-sign"), color = "green")
})
output$kpi.top5.gdp2 <- renderValueBox({
valueBox(paste("$", round(var_maxDate_kpi_top5_gdp()$gdpPercap[2]/1000,1), "k"),
paste(var_maxDate_kpi_top5_gdp()$country[2], " (", input$dateEnd, ")", sep = ""), icon = icon("dollar-sign"), color = "green")
})
output$kpi.top5.gdp3 <- renderValueBox({
valueBox(paste("$", round(var_maxDate_kpi_top5_gdp()$gdpPercap[3]/1000,1), "k"),
paste(var_maxDate_kpi_top5_gdp()$country[3], " (", input$dateEnd, ")", sep = ""), icon = icon("dollar-sign"), color = "green")
})
output$kpi.top5.gdp4 <- renderValueBox({
valueBox(paste("$", round(var_maxDate_kpi_top5_gdp()$gdpPercap[4]/1000,1), "k"),
paste(var_maxDate_kpi_top5_gdp()$country[4], " (", input$dateEnd, ")", sep = ""), icon = icon("dollar-sign"), color = "green")
})
output$kpi.top5.gdp5 <- renderValueBox({
valueBox(paste("$", round(var_maxDate_kpi_top5_gdp()$gdpPercap[5]/1000,1), "k"),
paste(var_maxDate_kpi_top5_gdp()$country[5], " (", input$dateEnd, ")", sep = ""), icon = icon("dollar-sign"), color = "green")
})
# GGPLOT - GDP Per Capita
output$lineplotgdp <- renderPlot({
ggplot(var_date_and_country(), aes(x = year, y = gdpPercap/1000, color = country)) +
geom_line(lwd = 1.5) +
theme_grey() +
labs(x = "Year",
y = "GDP Per Capita (000's)",
title = paste("GDP Per Capita Trend over time (", input$country1, " v ", input$country2, ")", sep = ""))
})
output$lineplotPopn2 <- renderPlot({
ggplot(var_date_and_country(), aes(x = year, y = (pop/10^6), color = country)) +
geom_line(lwd = 1.5) +
theme_grey() +
labs(x = "Year",
y = "Population (Millions)",
title = paste("Country Population in millions over time (", input$country1, " v ", input$country2, ")", sep = ""))
})
output$histGDP <- renderPlot({
ggplot(data, aes(x = gdpPercap, color = continent, fill = continent)) +
geom_histogram() +
theme_grey() +
labs(x = "GDP Per Capita",
y = "Frequency",
title = "GDP Per Capita distribution by Continent")
})
# Value Boxes - Bottom 5 KPIs | GDP Per Capita
output$kpi.btm5.gdp1 <- renderValueBox({
valueBox(paste("$", round(var_maxDate_kpi_btm5_gdp()$gdpPercap[1]/1000,1), "k"),
paste(var_maxDate_kpi_btm5_gdp()$country[1], " (", input$dateEnd, ")", sep = ""), icon = icon("dollar-sign"), color = "red")
})
output$kpi.btm5.gdp2 <- renderValueBox({
valueBox(paste("$", round(var_maxDate_kpi_btm5_gdp()$gdpPercap[2]/1000,1), "k"),
paste(var_maxDate_kpi_btm5_gdp()$country[2], " (", input$dateEnd, ")", sep = ""), icon = icon("dollar-sign"), color = "red")
})
output$kpi.btm5.gdp3 <- renderValueBox({
valueBox(paste("$", round(var_maxDate_kpi_btm5_gdp()$gdpPercap[3]/1000,1), "k"),
paste(var_maxDate_kpi_btm5_gdp()$country[3], " (", input$dateEnd, ")", sep = ""), icon = icon("dollar-sign"), color = "red")
})
output$kpi.btm5.gdp4 <- renderValueBox({
valueBox(paste("$", round(var_maxDate_kpi_btm5_gdp()$gdpPercap[4]/1000,1), "k"),
paste(var_maxDate_kpi_btm5_gdp()$country[4], " (", input$dateEnd, ")", sep = ""), icon = icon("dollar-sign"), color = "red")
})
output$kpi.btm5.gdp5 <- renderValueBox({
valueBox(paste("$", round(var_maxDate_kpi_btm5_gdp()$gdpPercap[5]/1000,1), "k"),
paste(var_maxDate_kpi_btm5_gdp()$country[5], " (", input$dateEnd, ")", sep = ""), icon = icon("dollar-sign"), color = "red")
})
# LINEAR MODELLING ----
# Format Predictor and Response col names
var_EDA_plots <- reactive({
str_col_predictor <- as.character(input$predictor)
str_col_response <- as.character(input$response)
transformed_data <- data %>% select(str_col_response, str_col_predictor, continent, country, year)
transformed_data$predictor_base <- unlist(transformed_data[2])
transformed_data$predictor_log <- unlist(log(transformed_data[2]))
transformed_data$predictor_inv <- unlist(1/transformed_data[2])
transformed_data$response <- unlist(transformed_data[1])
transformed_data %>% select(continent, country, year, response, predictor_base, predictor_log, predictor_inv)
})
# Build linear models reactively
var_lm_base <- reactive({
fml <- as.formula("response ~ predictor_base")
lm(fml, data = var_EDA_plots())
})
# Base EDA Scatter plot
output$EDAplotBase <- renderPlot({
ggplot(var_EDA_plots(), aes(x = predictor_base, y = response)) +
geom_point(aes(color = continent)) +
geom_smooth(method = "lm", se = TRUE) +
theme_grey() +
labs(x = paste(input$predictor, "(Predictor)"),
y = paste(input$response, "(Response)"),
title = paste(input$predictor, "v", input$response))
})
# Log EDA Scatter plot
output$EDAplotLog <- renderPlot({
ggplot(var_EDA_plots(), aes(x = predictor_log, y = response)) +
geom_point(aes(color = continent)) +
geom_smooth(method = "lm", se = TRUE) +
theme_grey() +
labs(x = paste("log(", input$predictor, ") (Predictor)", sep = ""),
y = paste(input$response, "(Response)"),
title = paste("log(", input$predictor, ") v ", input$response, sep = ""))
})
# Inv EDA Scatter plot
output$EDAplotInv <- renderPlot({
ggplot(var_EDA_plots(), aes(x = predictor_inv, y = response)) +
geom_point(aes(color = continent)) +
geom_smooth(method = "lm", se = TRUE) +
theme_grey() +
labs(x = paste("1/", input$predictor, " (Predictor)", sep = ""),
y = paste(input$response, "(Response)"),
title = paste("1/", input$predictor, " v ", input$response, sep = ""))
})
# Base LM Model
output$base.lm <- renderText({
summary(var_lm_base())
})
output$EDAtable <- renderTable({var_EDA_plots()})
}
shinyApp(ui, server)
Any help would be really appreciated, thanks :)
renderText() can't handle a list. Using summary with an lm object will result in a list of length 11. The easiest way to fix it is by changing the render function to renderPrint() instead.
# Base LM Model
output$base.lm <- renderPrint({
summary(var_lm_base())
})