search

RJT1990 / pyflux

Open source time series library for Python
BSD 3-Clause "New" or "Revised" License
2.1k stars 240 forks source link

Disagreement numbers on 'Normal Scale' #96

Closed hanhanzhai closed 7 years ago

hanhanzhai commented 7 years ago

Latent Variable Estimate Std Error z P>|z| 95% C.I.
Constant 0.0001 0.0001 0.8015 0.4228 (-0.0001 | 0.0002)
AR(1) 0.2983 0.1532 1.9476 0.0515 (-0.0019 | 0.5985)
MA(1) -0.4225 0.1454 -2.9064 0.0037 (-0.7075 | -0.1376)
Normal Scale 0.0059

None ['Constant', 'AR(1)', 'MA(1)', 'Normal Scale'] [ 6.35185457e-05 2.98320573e-01 -4.22547890e-01 -5.13708426e+00]

From above ARIMA fit summary and latent variable values, 'Normal Scale' are not the same. Is it by design? Thanks!

RJT1990 commented 7 years ago

Hey there - yes, the reason for this is that for the scale \sigma we transform exponentially and estimate theta instead, where sigma = exp(theta). So to get sigma, take exp(-5.137...) which should equal 0.0059. Hope that helps, Ross

hanhanzhai commented 7 years ago

I see. Thanks a lot for your explanation, Ross!