The SmootForecastApp OS X application provides easy to use time series forecasting capabilities. While sophisticated algorithms are used under the hood to provide forecasts, the user interface is intended to be simple so non-experts can access the capabilities. Some of the features provided are:
– Time series input can be from an existing text file or entered one line at a time by hand. The results can then be saved into a text file.
– Basic defaults so a forecast can be generated immediately. Help is provided with several complete examples with accompanying screen shots.
– Outlier filtering is provided with the ability to change the outlier filter window (which usually is some multiple of any seasonal cycle). But a general default window value is provided.
– Seasonal cycle can be specified or automatically generated if there is a detectable seasonal cycle.
– The most common trends are supported; linear and damped.
– An existing most recent portion of the time series can be held back and the remaining series used to generate a forecast for that period to see the effectiveness the current settings.
– Results are graphed in a simple display.
– Each aspect of the forecast process can be printed; from series input to series graph. It is up to the user to select which aspects are printed or which can be saved into PDF documents (if any).
