ENH: Add seasonal plotting. #1290

Merged
merged 6 commits into from Jan 9, 2014

Projects

None yet

3 participants

@jseabold
Member
jseabold commented Jan 5, 2014

Starting to add some seasonality stuff. Hoping to improve this area for 0.6.0. This is for seasonal plotting. See the __main__ script for examples.

@coveralls

Coverage Status

Coverage remained the same when pulling b9b86ba on jseabold:tsa-plots into f46421a on statsmodels:master.

@josef-pkt
Member

needs smoke test

@josef-pkt
Member

we also have the elnino dataset for monthly data, used as example in functional.fboxplot

@jseabold
Member
jseabold commented Jan 6, 2014

Yes, I'll add a test. Still a bit a WIP. I wanted to look once more at any additional bells and whistles. Will also probably need a little more robust error checking.

@josef-pkt
Member

related: I think we should eventually review our plots for consistent "style" keywords.

related: Two weeks ago I had problems finding the time series plot.
VAR has additional plots but only attached as methods. (I needed to fix the broken varplots file in the documentation.)

@jseabold
Member
jseabold commented Jan 6, 2014

To some extent this isn't possible, because matplotlib itself doesn't use consistent style keywords. There are also multiple plotting calls within each plotting function, so the style keywords will have to be replicated for each plot if we want to go the micro-optimization route. My current thinking is just to provide some quick plotting with minimal fussiness and a default style. If you want publication quality, you're going to have roll your own or do your style-tweaking ex post from the returned fig.

What broke with the varplots again? Issue number?

@josef-pkt
Member

What broke with the varplots again?

numpy string conversion of years used only the first character which messed up pandas, using dtype S4 fixed it.

@josef-pkt
Member

I think we need to advertise the plots better, (if mwaskom didn't even know we have any)

On the signature level, I thought we could have something like layout_kwds, plot_kwds, and what goes in there depends on the plot function and on how much users and developers want to fine tune specific plots.

@jseabold
Member
jseabold commented Jan 6, 2014

A gallery page would go a long way for advertising. Someone just needs to step up and do it. My involvement is going to be pretty minimal for the foreseeable future unless I need things for projects I have going.

Re: signature. That sounds fine, but in practice I haven't found it that simple. In any given plot, you may have a line, a scatter, some hlines, some vlines, etc. Do we provide vline1_kwargs, vline2_kwargs, plot_kwargs, etc. in the signature? Seems easier just to let people fix style ex post given it's all available from the fig.

@jseabold
Member
jseabold commented Jan 8, 2014

Rebased and added some smoke tests.

@coveralls

Coverage Status

Coverage remained the same when pulling 2cb7820 on jseabold:tsa-plots into f8156e7 on statsmodels:master.

@coveralls

Coverage Status

Coverage remained the same when pulling 3760606 on jseabold:tsa-plots into f8156e7 on statsmodels:master.

@jseabold
Member
jseabold commented Jan 9, 2014

Look ok to merge?

@coveralls

Coverage Status

Coverage remained the same when pulling 5b44dc0 on jseabold:tsa-plots into f8156e7 on statsmodels:master.

@jseabold jseabold merged commit 9d86d62 into statsmodels:master Jan 9, 2014

1 check passed

default The Travis CI build passed
Details
@jseabold jseabold deleted the jseabold:tsa-plots branch Jan 9, 2014
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment