Commit graph

15 commits

Author SHA1 Message Date
Andrew Johnson
2bbcd2c698
Use normal glm distribution for Stan model (#2051) 2022-05-24 22:11:22 +10:00
Brian Ward
9968f8be41
Simplify setup.py, make CmdStan the default backend (#2088) 2022-04-05 00:06:59 +10:00
Ryan Nazareth
ac59b44ca3
add implementation for constant trend in Python (#1466)
* add implementation for constant trend

* force k and delta params to be 0s

* add tests and fix n_changepoints, changepoints_t to 0

* Add test for cv with constant trend

* Add docs and test for checking invalid input

* make changes to stan

* add transformed params block in stan and output flat trend vector

* correct syntax

* transformed params syntax

* Fix test and port changes to win stan file

* add test for flat trend function

* Add separate function for flat trend init

* fix test
2020-05-14 21:40:40 -07:00
Ben Letham
7ff33137eb Change to MIT license 2019-05-21 11:40:04 -07:00
bl
a38aaa38c6 Stan fix for pystan 2.16 2018-05-30 14:23:37 -07:00
Ben Letham
41d4ab472f Set noise prior variance to current value for linear growth model 2018-05-29 14:41:36 -07:00
Ben Letham
663214393e Minor stan simplifications 2018-05-25 13:57:45 -07:00
Ben Letham
aa37fb48ac Fix Stan file for Windows 2018-05-22 09:27:00 -07:00
Ben Letham
3bd372bc15 Speed up stan fitting by removing unecessary parameter definitions 2018-05-15 13:26:46 -07:00
Ben Letham
8d8c5b41ce Multiplicative seasonality (Py) 2018-05-09 12:25:29 -07:00
Ben Letham
55d7d1e62d Single stan model with both trends (Py) 2018-05-04 16:04:29 -07:00
bl
8f1607cd93 Extra regressors Py 2017-07-21 07:05:16 -07:00
Rolando (Max) Espinoza
e33e7c4b37 Make stan code windows-compatible. (#96)
The vector/matrix operations fail to compile in windows due to eigen
incompatibility with windows compiler. Here we opt to use a non-
vectorized model for windows platform.
2017-03-09 18:47:59 +02:00
Ben Letham
443d475468 Allow changepoints on dates that aren't in history, and allow for repeated observations on days. Previously we worked with changepoints via their index in the history. Now we work with them using just their value in scaled time. 2017-02-28 10:08:37 +02:00
Sean J. Taylor
e51b42b336 Initial commit 2017-02-22 15:59:43 -08:00