Prognos

Prognos is a forecasting libray in Scala similar to R forecast package. It's implemented using breeze numerical computing library.

Algorithms in this library can be used from Scalding, Spark and other Java/Scala applications. Check tests, prognos-scalding-example and prognos-spark-example for usage examples.

Forecasting algorithms in this library are based on the book Forecasting Principles and Practice by Rob Hyndman

For using it from a scala application, add this as dependency in build.sbt, after publishing it to local/artifact repository:

libraryDependencies += "com.fk" % "prognos_2.10" % "1.0"

Forecasting algorithms

Available forecasting algorithms:

• Moving Average
• Simple Exponential Smoothing
• Holt Linear and Exponential
• HoltWinters

[TODO] Forecasting algorithms

Please send a pull request for other algorithms:

• Holt Damped Linear/Exponential
• Optimization to find alpha, beta, gamma parameters in Holt's method
• ARIMA
• Auto ARIMA

How to contribute

1. Add timeseries data from R fpp package or other sources to testdata package
2. Add unit test for validating forecast output
3. Add forecasting algorithm
4. Send a pull request
5. For publishing locally and testing, use $ sbt publish-local