library(shiny)
library(mlbench)
library(ggplot2)
library(dplyr)
library(tidyr)
library(class)
library(caret)
library(corrplot)
library(boot)
library(randomForest)
library(gbm)

ui <- fluidPage(
  titlePanel("Classification and clustering"),
  sidebarLayout(
    sidebarPanel(
      selectInput(
        'select_predictors',
        label = 'Define predictors',
        choices = c('RI', 'Na', 'Mg', 'Al', 'Si', 'Ca', 'K', 'Ba', 'Fe'),
        selected = c('RI', 'Na', 'Mg', 'Al', 'Si', 'Ca', 'K', 'Ba', 'Fe'),
        multiple = TRUE
      )
    ),
    mainPanel(
      tabsetPanel(
        tabPanel(
          title = "tab 1",
          value = "tab1",
          verbatimTextOutput('general_linear_model')
        ),
        tabPanel(
          title = "tab 2",
          value = "tab2",
          plotOutput("cor_plot")
        ),
        tabPanel(
          title = "tab 3",
          value = "tab3",
          verbatimTextOutput('summary_glass_ds')
        )
      )
    )
  )
)

server <- function(input, output) {
  data(Glass)
  df_glass <- Glass[c(1:146),]
  df_glass$Type <-factor(df_glass$Type)

  output$summary_glass_ds <- renderPrint({
    summary(df_glass)
  })

  #change dataset_type for corplot

  df_num <- df_glass
  df_num$Type <- as.numeric(df_num$Type)
  df_num$Type[df_num$Type==1] <- 0
  df_num$Type[df_num$Type==2] <- 1
  cor_tab <- cor(x=df_num[])

  output$cor_plot <- renderPlot({
    corrplot(cor_tab, 'number')
  })


  output$general_linear_model <- renderPrint({
    measurevar <- "Type"
    groupvars  <- input$select_predictors

    gen_lin_mod <- glm(data=df_num, as.formula(paste(measurevar, paste(groupvars, collapse=" + "), sep=" ~ ")), family = 'binomial')
    summary(gen_lin_mod)
    # paste(measurevar, paste(groupvars, collapse=" + "), sep=" ~ ")
  })

}

# Return a Shiny app object
shinyApp(ui = ui, server = server)