0.5.0

Added params features to make it easier to see what was fit.

0.4.9

update versions of deps

0.4.8

Update for recent Python installs and newer pytest

0.4.6

0.4.6 Adding TimeSeriesInitialSplit

In this update I added TimeSeriesInitialSplit. This will split the data in a convenient way for time series testing.

0.4.5

Droping Stan

I love Stan, but I had to drop it as the backend sampler.

A lot of problems were solved as this package was developed. Multiprocessing
with cmdstanpy folder structures was a big one. As I used this package more
in production I started to dread it more than love it. Why? A lot of small
simple things. Keeping tabs on cmdstanpy, getting a version of python
that all of the dependancies were happy with felt tedious. In prod environments
you also have to figure out if you have C++ compilation tools like gcc.
This is not always the case. If those tools are not available then install_cmdstanpy
will not work. This is additional overhead, and something I did not want to
have to worry about. Once I got
a python version 3.8>= to work well, I found that when I wanted to make
a notebook example with Google Colab I had to wrestle with the notebook
to get a more recent version of python -- Colab comes with 3.7 by default.
The version was not all though. I had to make sure that the priority was
reset to run the more recent version. This was not the end. I also had
to install additional software that comes with the newer version of
python, but does not come with the notebook -- something called dsutils?

Why would I care so much about the notebook environment? Well, I think
there are a population of people who like to run code on notebooks. This
could be how they 'kick the tires' when trying a new package. If that
experience goes well then maybe the chances of making into production is
higher.

Though I love Stan, and it hurts to remove it, I think that these models
are simple enough to survive just fine without stans sampler.

If you are interested in how to use stan with a python then checkout out
version 4.4 of this package commit hash 49066323c08915e90b8b1e06a88cbddb52a726ae

I think these changes will make the package easier for me to maintain and
very simple to integrate for production runs.

What's Changed

• Create LICENSE by @alexhallam in #1
• refactor and clean by @alexhallam in #2
• make frequency a required parameter. Update tests and docs by @alexhallam in #4
• Update snaive.stan by @alexhallam in #5
• add build for pypi by @alexhallam in #8
• testing chain_id by @alexhallam in #9
• working Pool by @alexhallam in #10
• update snaive to make directory structures that do not conflict in mu… by @alexhallam in #12
• update for website and clean code a little by @alexhallam in #15
• rename www to docs for github page by @alexhallam in #16
• add web updates to github yaml by @alexhallam in #17
• debug liquid layout by @alexhallam in #18
• update github workflow by @alexhallam in #19
• Website code update by @alexhallam in #20
• more yaml work by @alexhallam in #21
• Website code update by @alexhallam in #22
• more docs work by @alexhallam in #23
• Clean/green by @alexhallam in #24
• try more work on mkdocstring by @alexhallam in #25
• add deploy in yml by @alexhallam in #26
• add docs by @alexhallam in #27
• clean api by @alexhallam in #28
• need to add installs by @alexhallam in #29
• add pages by @alexhallam in #30
• clean readme by @alexhallam in #31
• add skill score example to docs by @alexhallam in #32
• add some ts data by @alexhallam in #33
• Bug/local/multiproc by @alexhallam in #34
• fix docs index by @alexhallam in #35
• bump version by @alexhallam in #36
• update pypi links by @alexhallam in #37
• Feature/allow alt run by @alexhallam in #39
• Features/clean no stan run by @alexhallam in #40
• Update pyproject.toml by @alexhallam in #42
• clean snaive by @alexhallam in #43
• Non stan by @alexhallam in #44

New Contributors

• @alexhallam made their first contribution in #1

Full Changelog: https://github.com/alexhallam/tablespoon/commits/0.4.5