There are three main sections of the application. The left-hand bar allows users to select the product of interest, the forecasting model to be used, and finally, the metric to forecast on (quantity and revenue):

The preceding forecasting models are supported (Auto, Holt—Winters, TBATS, Auto ARIMA, and Markov Chain Monte-Carlo). The forecasting metrics are as follows:

The main screen shows a chart that compares the metric of interest across the top five products in the same category as the product selected in the left-hand menu. The chart has been plotted using the Plotly API available from: https://plot.ly/r/:

The second half of the screen shows the plots of the actual data versus the predicted value, with the respective confidence intervals shown in the shaded areas in the first plot image.

The right side of the section lists the forecasted values, along with the 80% and 95% confidence intervals:

The third and final section shows the actual data in tabular format. Note that the data is aggregated by month and the columns show the actual cost, the quantity, and other related information:

One of the options in the forecasting model menu includes MCMC, which is a more sophisticated forecasting method available via the prophet package in R. It was released by Facebook and contains advanced forecasting capabilities. That said, practitioners tend to rely more heavily on time-tested techniques such as Holt—Winters and ARIMA, both of which have also been included as options in the drop-down menus.