Material Detail
Which forecast for which signal? Variability, memory and forecastability
Is there a best forecast? On synthetic signals (random, correlated, Brownian or cyclic), learners change the memory, noise and horizon with sliders and see which simple forecast wins: the mean, persistence, cyclic persistence, their convex combinations (CLIPER, BLEND) or a two-lag autoregression. On a random signal the mean wins; on Brownian motion persistence wins; with a cycle the result depends on the horizon. The page introduces...
Show MoreQuality
- User Rating
- Comments
- Learning Exercises
- Bookmark Collections
- Course ePortfolios
- Accessibility Info