Shiny application slow

[daniel_stockholm]

My shiny application (curve fitting) is very slow compare to running in R script. >10times slower . How can I improve?

Server code

library(shiny)
library(minpack.lm)
library(deSolve)
library(ggplot2)
library(FME)

shinyServer(function(input, output) {
#get datafile
  datfile<-reactive({
    req(input$data_att)
    
    dataspec<-input$data_att
    aa<-read.table(dataspec$datapath,header=TRUE, sep=",")
    return(aa)
    
  })

  #initiate curve fitting with button "Fitting"
  curvefitting<-eventReactive(input$Fitting,{
   #initial parameters"
     parms<-c(ka=1E4,kd=1E-4,bmax=40)
  
       Fit<-modFit(cost,p=parms)
    ab<-coef(Fit)
  })
  
 
   output$test<-renderText({curvefitting()})
  
   
  
  cost<-function(parms){
    
    out<-ode(y<-c(F=0),times=seq(0,2406,1),model,parms)
    dataout<-as.data.frame(out)
    modCost(dataout,datfile())
  }
 # model, Import function due to signal vary in time,
  model <- function (time, y,parms) {
    with(as.list(c(y,parms)), {
     
      import<-signal(time)
      dF <- (ka*import*(bmax-F) -kd*F)
      
      
      list(dF)
    })
  }
 
  signal <- function(time) {
    times <- seq(0, 2406, 1)
    
    ca0<-rep(0,15)
    ca1<-rep(1E-7,210)
    ca2<-rep(0 ,158)
    ca3<-rep(2E-7 ,210)
    ca4<-rep(0 ,158)
    ca5<-rep(4E-7 ,210)
    ca6<-rep(0,158)
    ca7<-rep(8E-7 ,210)
    ca8<-rep(0,158)
    ca9<-rep(16E-7 ,210)
    ca10<-rep(0,710)
    
    concentrations<-list(ca0,ca1,ca2,ca3,ca4,ca5,ca6,ca7,ca8,ca9,ca10)
    conc<-Reduce(c,concentrations)
    signal <- data.frame(times, conc)
    import <- approxfun(signal, rule = 1)
    import(time)
  }
 
   

  #end of program  
})

UI

library(shiny)

# 
shinyUI(fluidPage(
  
    titlePanel("Single Cycle"),

   sidebarLayout(
     sidebarPanel(

  fileInput("data_att","upload the file"),

  textOutput("test"),
 
  actionButton("Fitting", "Run Curve fitting")
  ),

         #not used
        mainPanel(plotOutput("dissanalys"))
                    
                 
    )
)
)

[nirgrahamuk]

generally speaking the first step to addressing a slow code issue is to profile the code, to understand what the slow parts are. I use package profvis for profiling.

I must say it does seem very unlikely that doing a like for like calculation in shiny should be measurably slowering than doing it in a plain script on the same device ...