Review of time series code during development

[astrofrog]

About 📈

This PR is to show the current progress in the development of the proposed astropy.timeseries package based on the (still in progress) Astropy Proposal for Enhancements 9.

If you are interested in time series, please feel free to take a look at this PR, read the proposed documentation and try out the code - please do feel free to open pull requests against this branch if you would like to contribute any fixes/improvements!

Once there is a general consensus that this is on the right track, and has been polished up, a PR will be opened to the core astropy repository for a wider review.

Current status 🕐

The two main base classes (for sampled and binned time series) are in place, and we have some basic functionality such as stacking, resampling, and folding working already. The documentation is in reasonably good shape and is ready for some reviews.

Documentation 📖

If you are interested in reading the documentation, here is a preview - we could really do with some reviews to make sure we are on the right track!

The documentation is actually the most important aspect to review at the moment since the details of the implementation shouldn't matter (we can always optimize performance issues later). The documentation also acts as tests, so the functionality shown in the documentation works.

Things that are missing/incomplete currently 🚧

If you'd like to help with these, please leave a comment!

• [ ] Some kind of interpolation/upsample functionality - it's not clear to me exactly how this should work, so it would be good to hear from people who actually need this.
• [ ] Documentation about masking - this can be short, since it's basically the same as for Table, so we can just link to that.
• [ ] Functionality for the binned time series - at the moment the folding and resampling are for the SampledTimeSeries class - the BinnedTimeSeries class doesn't really have any analysis functionality available
• [ ] We should make the periodogram functionality in astropy.stats aware of the time series classes, and possibly even consider moving that functionality to astropy.timeseries
• [ ] We should add support for more time series file formats, if there are any other standard ones
• [ ] The Kepler/TESS readers may be able to share some code since the files are similar, and the readers as they are written may be incomplete, so it'd be good to get input from experts
• [ ] Better integration with Matplotlib, specifically the ability to show times on one of the axes
• [ ] Consider writing a Time sub-class that lazily evaluates evenly spaced times - this is maybe a longer term optimization and not needed immediately
• [ ] Rename base TimeSeries class to BaseTimeSeries and consider whether SampledTimeSeries should simply be TimeSeries.
• [x] Implement proper support for to_pandas()/from_pandas()
• [ ] Investigate whether we can write a generic reader that can be told which column is the time and what format it is in
• [ ] For binned time series, determine how loc should work
• [ ] Make it possible to use loc with strings and without having to pass a Time object

[codecov-io]

Codecov Report

Merging #2 into master will decrease coverage by <.01%. The diff coverage is 87.34%.

@@            Coverage Diff             @@
##           master       #2      +/-   ##
==========================================
- Coverage   86.92%   86.91%   -0.01%     
==========================================
  Files         383      383              
  Lines       57823    57855      +32     
  Branches     1056     1056              
==========================================
+ Hits        50264    50287      +23     
- Misses       6945     6954       +9     
  Partials      614      614

Impacted Files	Coverage Δ
astropy/table/table.py	92.93% <ø> (ø)	:arrow_up:
astropy/nddata/ccddata.py	93.67% <ø> (ø)	:arrow_up:
astropy/stats/bls/core.py	90.99% <ø> (ø)	:arrow_up:
astropy/wcs/src/wcslib_wrap.c	65.67% <ø> (ø)	:arrow_up:
astropy/convolution/kernels.py	97.38% <100%> (ø)	:arrow_up:
astropy/visualization/hist.py	100% <100%> (ø)	:arrow_up:
astropy/stats/bls/methods.py	100% <100%> (ø)	:arrow_up:
astropy/utils/xml/src/iterparse.c	62.25% <100%> (ø)	:arrow_up:
astropy/io/votable/src/tablewriter.c	60.6% <100%> (ø)	:arrow_up:
astropy/stats/bls/bls.c	85.88% <100%> (+0.69%)	:arrow_up:
... and 6 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 0725a5c...135a94a. Read the comment docs.

[astrofrog]

Closing since astropy.timeseries is now in astropy