← All widgets
Kalman Filter / Local Level Explorer
ECON-5371 · Time Series Analysis and Forecasting — Chapter 11 companion widget
Local Level Model
y
t
= μ
t
+ ε
t
μ
t
= μ
t−1
+ η
t
Noise Variances
σ²
ε
(observation noise)
4.0
σ²
η
(state noise)
0.5
Signal-to-Noise Ratio
q = σ²
η
/ σ²
ε
0.125
Steady-state Kalman gain
—
Display
Show true (hidden) trend
Show noisy observations
Show smoothed estimate
Reshuffle draw
Reset
Filtered vs. Smoothed Trend Estimates
Observed y(t)
True trend μ(t)
Filtered μ̂(t|t)
Smoothed μ̂(t|T)
Kalman Gain K(t) Over Time