Forecasting Using Time Series Analysis – SAP Predictive Expert Analytics

Using data from the Bureau of Transportation Statistics ( http://www.rita.dot.gov/bts/acts ) I downloaded US domestic passenger data by year/month from 2005 to January 2015.

Passenger airline data (at least in the US) is seasonal, with high volumes in the summer months.  Given a series of actual / historical data values we can use triple exponential smoothing algorithms to project the data. The airline passenger data are a series of data points over time, giving the phrase time series analysis.

I am using SAP Predictive Expert Analytics.  In the above figure, I have loaded the .CSV file of airline passenger data.

I select R-triple exponential smoothing, because as it shows above it provides “seasonality based time series forecasting”

See R – Forecasting – Training Material

I can drag it on to the workspace.

The above shows I select forecast mode as I want to forecast the data.

The dependent column is the column we wish to forecast, in this case, passenger count

Period for the data is monthly

After running the model, I see success and switch to the Results view.

Above shows the forecast passenger values by month for 2015 and part of 2016.

For fun I also run the SAP-provided triple exponential smoothing (not the R-algorithm)

It provides similar forecast results, with July 2015 being the high month of projected passenger travel in the US.

Reference:

Helping you Predict Your Future – ASUG Annual Conference & Other Upcoming Items of Interest