Spotfire Tips & Tricks: Single Click Forecasting with Spotfire

Introduction

Forecasting is a process to make future predictions based on past and future data. There are Qualitative-forecasting methods like the Delphi method relies on a panel of experts. Quantitative forecasting methods forecast the future as a function of past data.

In virtually every decision everyone uses some kind of forecast. Starting from what to wear as a function of the weather forecast. Sound predictions of demands and trends are no longer for Geeks, but a necessity for everyone. If organizations have to cope with seasonality, sudden changes, maneuvers of the competition, rapid technology changes, and large swings in the economy- Forecasting can turn out as a handy tool. It may still be driving forward while viewing the road through the rear view window and that makes risk and uncertainty part of the process.

Holt-Winters Forecast

This is one of the popular methods when data exhibits trend and seasonality.  It is one of the Exponential Smoothing methods. The two main HW models are the Additive model for time series exhibiting additive structure and the Multiplicative model for time series exhibiting Multiplicative structure

Single Click Forecast in Spotfire

The Holt-Winters Forecast uses TERR (Spotfire Enterprise Runtime for R) to compute the Holt-Winters filtering of a time series or anything that can be coerced to a time series. This is an important requirement that translates as visualization having X axis as Time series data.

This is an exponentially weighted moving average filter of the level, trend, and seasonal components of a time series. The smoothing parameters are chosen to minimize the sum of the squared one-step-ahead prediction errors.

Adding Holt-Winters Forecast Curve in Spotfire

It is simple to add curves by either right click Forecast or under visualization, properties selecting Lines & curves and then add Forecast-Holt Winters. The output of a Holt-Winters Forecast is three different curves: a fitted curve showing the general variation of the measure of interest, a forecast curve predicting the future trend, and a confidence interval showing how the insecurity increases the further away from the known values the prediction reaches.

The parameters are automatically calculated but they can be edited as well. It is also possible to have a curve for each Color, Line, or Trellis Panel.

The Spotfire user guide provides a detailed description of each parameter  

At times curve may not show up ? the user will be informed with a detailed error message specifying the reason like in the example below multiple empty values in succession are the reason you do not see the Curve. It is not only Line charts but Forecast curves that can be added to other visualizations like Scatter plots.

Export Forecast Data as Table

IronPython scripting can be used to export the forecast fitted curve to data table. Follow this tutorial to see how.