Reputation: 545
I want to build forecast techniques that exponential smoothing method is one of my selection. However, I have some issues with representing the ggplot and the result/report of the calculation.
Initially, I am generating random dataset in order to be used for this technique where alpha and number of periods to be forecasted are determined by the user. For instance; i have 100 days and next 4 days are willing to be estimated with their lines -fit, upper and lower-. Then I want to learn the values of this data as a table.
When I try to visualize the plot, the error is: ggplot2 doesn't know how to deal with data of class mtstsmatrix
Secondly, I would like to monitor the data like:
require(shiny)
require(ggplot2)
require(forecast)
require(TTR)
shinyServer(function(input, output, session){
set.seed(123)
output$es1 <- renderPlot({
tmp <- data.frame(time = 1:100, sales = round(runif(100, 150, 879)) )
tmp.mean <- HoltWinters(x=tmp$sales, alpha = input$alpha, beta = FALSE,gamma=FALSE)
tmp.pred <- predict(tmp.mean,n.ahead = input$h, prediction.interval = TRUE)
y <- ggplot(tmp, aes(time, sales)) +
geom_line() +
geom_line(data=tmp.pred, aes(y=tmp.pred[,1]),color="red") +
geom_line(data=tmp.pred, aes(y=tmp.pred[,2]),color="blue") +
xlab("Days") +
ylab("Sales Quantity")+
ggtitle(title)
y })
output$infoes <- renderDataTable({
tmp <- data.frame(time = 1:100, sales = round(runif(100, 150, 879)) )
tmp.mean <- HoltWinters(x=tmp$sales, alpha = input$alpha, beta = FALSE,gamma=FALSE)
tmp.pred <- predict(tmp.mean,n.ahead = input$h, prediction.interval = TRUE)
tmp.pred
})
ui
require(shiny)
require(ggplot2)
require(forecast)
require(TTR)
shinyUI(pageWithSidebar(
headerPanel("Forecasting Methods"),
sidebarPanel(
h3(strong("Exponential Smoothing",style = "color:black")),
br(),
sliderInput("h","Number of periods for forecasting:",
min = 1, max = 20, step= 1, value = 4),
sliderInput("alpha","Alpha (Smoothing Parameter):",
min = 0.05, max = 1, step= 0.05, value = 0.01)
),
mainPanel(
tabsetPanel( id="tabs",
tabPanel("Exponential Smoothing",
value="panel",
plotOutput(outputId = "es1",
width = "900px",height = "400px"),
dataTableOutput(outputId="infoes"))
))))
Upvotes: 0
Views: 947
Reputation: 6913
You had a to make tmp.pred
palatable for ggplot
as was said in the comments. You also don't have to create the same data in multiple statements, a reactive
command is good for that:
ui.R (unchanged)
require(shiny)
require(ggplot2)
require(forecast)
require(TTR)
shinyUI(pageWithSidebar(
headerPanel("Forecasting Methods"),
sidebarPanel(
h3(strong("Exponential Smoothing",style = "color:black")),
br(),
sliderInput("h","Number of periods for forecasting:",
min = 1, max = 20, step= 1, value = 4),
sliderInput("alpha","Alpha (Smoothing Parameter):",
min = 0.05, max = 1, step= 0.05, value = 0.01)
),
mainPanel(
tabsetPanel( id="tabs",
tabPanel("Exponential Smoothing",
value="panel",
plotOutput(outputId = "es1",
width = "900px",height = "400px"),
dataTableOutput(outputId="infoes"))
))))
server.R
require(shiny)
require(ggplot2)
require(forecast)
require(TTR)
shinyServer(function(input, output, session){
set.seed(123)
predset <- reactive({
tmp <- data.frame(time = 1:100, sales = round(runif(100, 150, 879)) )
tmp.mean <- HoltWinters(x=tmp$sales, alpha = input$alpha, beta = FALSE,gamma=FALSE)
tmp.pred <- data.frame(predict(tmp.mean,n.ahead = input$h, prediction.interval = TRUE), time = tmp[nrow(tmp), "time"] + 1:input$h)
list(tmp = tmp, tmp.pred = tmp.pred)
})
output$es1 <- renderPlot({
tmp <- predset()$tmp
tmp.pred <- predset()$tmp.pred
y <- ggplot(tmp, aes(time, sales)) +
geom_line() +
geom_line(data=tmp.pred, aes(y=upr),color="red") +
geom_line(data=tmp.pred, aes(y=fit),color="blue") +
geom_line(data=tmp.pred, aes(y=lwr),color="red") +
xlab("Days") +
ylab("Sales Quantity")+
ggtitle("title")
y })
output$infoes <- renderDataTable({
predset()$tmp.pred
})
})
Upvotes: 1