Automatic model selection

hawkular / hawkular-datamining

Real-time time series prediction library with standalone server

36 stars 22 forks source link

Closed pavolloffay closed 8 years ago

pavolloffay commented 8 years ago

State is initialized as follows:

double exponential smoothing - simple: (level0=y0, slope0=y1-y0), complex: fit of linear regression
simple exponential smoothing - simple: (level0=y0), complex: mean(initData$y)

This implementations of exp. smoothing give lower MSE than R's ets(). Model selection and parameters estimation is much faster than OpenForecast

pavolloffay commented 8 years ago

Tests for continual learning needs to be added, then it is ready to merge.

pavolloffay commented 8 years ago

@Jiri-Kremser could you please review?