Forecasting has been an essential task in business operations and has become even more important in today's fast-changing times.

In order to understand the behavior of different traditional and more advanced forecasting techniques it is good starting point for the endusers to try them out directly.

This app allows the interactive assesment of eight different forecasting techniques for three different academic data sets.

The app runs online, thus, can be easily accessed even by non-technical users.

A very nice starting point on topic of forecasting is given in the online book by Rob J Hyndman and George Athanasopoulos which also has it's own R package fpp. In the app the user can choose from the following techniques that rely on the forecast and fpp packages:

• Naiive
• Mean
• Drift method
• Seasonal naiive
• Exponential smoothing
• ARIMA
• ETS
• Neural net

Left side - Adjustments

On the left side the user can choose from three different data sets as well as selecting the forecasting technique. Additionally, the horizon to forecast and the starting point of the forecast can be selected. Notice that the years are automatically adjusted depending on the selected data set.

Right side - Resulting output

The right side shows some explanatory text and the final result wich consists of the chosen time series (black) and the computed forecast (blue). For most (except neural net) also the confidence intervals of the forecast are given.

One can observe that with dates farther in the future also the confidence intervals are increasing.