two algorithms for time series forecasting
play

Two Algorithms for Time Series Forecasting Danny Yuan Forecasting - PowerPoint PPT Presentation

Two Algorithms for Time Series Forecasting Danny Yuan Forecasting with Fast Fourier Transformation Key Idea: Decomposition A reasonably continuous and periodic function can be expressed as the sum of a series of sine terms FFT Is Simple 1.


  1. Two Algorithms for Time Series Forecasting Danny Yuan

  2. Forecasting with Fast Fourier Transformation

  3. Key Idea: Decomposition

  4. A reasonably continuous and periodic function can be expressed as the sum of a series of sine terms

  5. FFT Is Simple 1. Run FFT on input data 2. Filter out low-amplitude, high-frequency components 3. Forecast on each individual component 4. Run inverse of FFT of filtered data 5. Profit!

  6. FFT Is Simple

  7. Solution: Iteratively Compensate Input with Error

  8. When Should We Use FFT?

  9. When There Is Periodicity

  10. When You Need a Quick Job

  11. Decomposition Is Powerful Reference

  12. Decomposition Is Powerful

  13. Decomposition Is Powerful

  14. Decomposition Is Powerful

  15. Where Is The Bottleneck? Not easy to combine new signals ● E.g. events, weather ●

  16. Forecasting With Deep Learning

  17. Key Idea: Time Series Are Sequences Time series can be discretized into sequence ● We can apply techniques of seq2seq ●

  18. Forecast Forecast Forecast Forecast Forecast 1 2 3 4 n H_0 H_1 H_2 H_3 H_n Input Input Start Input1 Input2 3 n-1 T0 T1 T2 T3 Tn t

  19. Forecast Forecast Forecast Forecast Forecast 1 2 3 4 n H_0 H_1 H_2 H_3 H_n Input Input Start Input1 Input2 3 n-1 …... [Forecast 1, Time of Week 1] [F_(n-1), TOW_(n-1)] t

  20. Forecast Forecast Forecast Forecast Forecast 1 2 3 4 n H_0 H_1 H_2 H_3 H_n Input Input Start Input1 Input2 3 n-1 …... [F_(n-1), TOW_(n-1), [Forecast 1, Time of Week 1, Weather_(n-1), X_(n-1)] Weather 1] [Temperature, Humidity, Precipitation, Wind, t Weather Type]

  21. What About Recent Context?

  22. Forecast Forecast Forecast Forecast Forecast 1 2 3 4 n h_1 h_2 h_3 h_m H_0 H_1 H_2 H_3 H_n Input Input Input Input Input Input Input Input Start 1 2 3 m 1 2 3 n-1 t - m t - m + 1 t - m + 2 t - 1 t …... [F_(n-1), TOW_(n-1), [F_1, TOW_1, Weather_1, X_1] Weather_(n-1), X_(n-1)]

  23. Encoder Decoder Forecast Forecast Forecast Forecast Forecast 1 2 3 4 n h_1 h_2 h_3 h_m H_0 H_1 H_2 H_3 H_n Input Input Input Input Input Input Input Input Start 1 2 3 m 1 2 3 n-1 t - m t - m + 1 t - m + 2 t - 1 t …... [F_(n-1), TOW_(n-1), [F_1, TOW_1, Weather_1, X_1] Weather_(n-1), X_(n-1)]

  24. Summary Decomposition is a powerful tool in time series forecasting ● Time series forecasting can be modeled as a seq2seq problem ●

Recommend


More recommend