In this small project we develop a shiny app that uses four time series forecasting methods to predict Australian quarterly beer production.

In this shiny app, you can select the forecasting method and accordingly the plot and table of real and predicted values are presented.

Description

Total quarterly beer production in Australia (in megalitres) from 1956:Q1 to 2010:Q2.

Format

Quarterly time series of class ts.

Source

Australian Bureau of Statistics. Cat. 8301.0.55.001.

Average method: forecasts of future values are just the average of the historical data.

Naïve method: we simply set all forecasts to be the value of the last observation

Seasonal naïve method: A similar method is useful for highly seasonal data. In this case, forecast is set to the last observed value from the same season of the year.

ARIMA models: ARIMA models are advanced models, more details are available HERE

forecasting results from Shiny app using ARIMA Model

pred <- forecast(auto.arima(beer2), h=6)
data <- cbind(as.numeric(real),as.data.frame(pred)[,c(1,4,5)])
colnames(data) <- c("Real Values", "Point Forecast", "Lower 95% CI", "Upper 95% CI" )
kable(data)