Page 161 - Contributed Paper Session (CPS) - Volume 4

CPS2164 Jonathan Hosking et al.
Table 1. Computation times for the models discussed in Section 6. “Analytical” computations
use analytical derivatives; “Numerical” computations use numerical derivatives; “Ratio” is the
ratio of the computation times for numerical and analytical derivatives.

derivatives using ana anlaytical expression rather than a finite-difference
approximation. This speeds up the computations by a factor that in practical
applications can be between 3 and 80.

References
1. Anderson, E. W., McGrattan, E. R., Hansen, L. P., and Sargent, T. J. (1995).
On the mechanics of forming and estimating dynamic linear economies.
In Handbook of Computational Economics, vol. 1, eds. H. M. Amman, D.
A. Kendrick, and J. Rust, pp. 171–252.
2. Ansley, C. F. and Kohn, R. (1985). A structured state space approach to
computing the likelihood of an ARIMA process and its derivatives. J.
Statist. Comput. Simul., 21, 135–169.
3. Durbin, J., and Koopman, S. J. (2012). Time Series Analysis by State Space
Methods. Oxford University
4. Press.
5. Hamilton, J. D. (1994). Time Series Analysis. Princeton University Press.
6. Harvey, A. C. (1989). Forecasting, Structural Time Series Models and the
Kalman Filter. Cambridge University Press.
7. Hosking, J. R. M., Natarajan, R., Ghosh, S., Subramanian, S., and Zhang, X.
(2013). Short-term forecasting of the daily load curve for residential
electricity usage in the Smart Grid. Appl. Stoch. Models Bus. Ind., 29,
604–620.
8. Hyndman, R. J., Koehler, A. B., Snyder, R. D., and Grose, S. (2002). A state
space framework for automatic forecasting using exponential smoothing
methods. Int. J. Forecasting, 18, 439-454.
9. Koopman, S. J., and Shephard, N. (1992). Exact score for time series
models in state space form. Biometrika, 79, 823–826.
10. Melard, G. (1985). Exact derivatives of the likelihood of ARMA processes.
In´ Proceedings of the Statistical Computing Section, pp. 187–192.
American Statistical Association, Washington, DC.

150 | I S I W S C 2 0 1 9