Exponentially weighted estimands and the exponential family: filtering, prediction and smoothing
By: Simon Donker van Heel, Neil Shephard
We propose using a discounted version of a convex combination of the log-likelihood with the corresponding expected log-likelihood such that when they are maximized they yield a filter, predictor and smoother for time series. This paper then focuses on working out the implications of this in the case of the canonical exponential family. The results are simple exact filters, predictors and smoothers with linear recursions. A theory for these models is developed and the models are illustrated on simulated and real data.
Similar Papers
Asymptotic Expansion for Nonlinear Filtering in the Small System Noise Regime
Signal Processing
Makes computer guesses about things more accurate.
Rates of Convergence of Maximum Smoothed Log-Likelihood Estimators for Semi-Parametric Multivariate Mixtures
Statistics Theory
Makes smart guesses about mixed data more reliable.
Maximum likelihood estimation for the $λ$-exponential family
Statistics Theory
Helps computers learn from data more accurately.