[WIP] Added TabularRegressionDataFrameDataSource wrapping TimeSeriesDataSet and TabularRegressionPreprocess stub

[sumanmichael]

What does this PR do?

Created TabularRegressionPreprocess, TabularRegressionDataFrameDataSource wrapping TimeSeriesDataSet in pytorch-forecasting

Here's what I did.

• Create TabularRegressionPreprocess stub
• created TabularClassificationPreprocess in new tabular/classification/data.py & TabularRegressionPreprocess in new tabular/regression/data.py extending the TabularPreprocess in tabular/data.py by fixing the respective is_regression bool param.
• created TabularClassificationDataFrameDataSource in new tabular/classification/data.py & TabularRegressionDataFrameDataSource in new tabular/regression/data.py extending the TabularPreprocess in tabular/data.py by wrapping TimeSeriesDataSet in pytorch-forecasting.
  • TabularRegressionData.from_data_frame(train_data, …, preprocess_kwargs)
  • that calls DataModule.from_data_source(“data_frame”, …)
  • that creates the TabularRegressionPreprocess with the preprocess_kwargs (including the args for the time series data set)
  • then from_data_source gets the data_frame data source and calls load_data with the data frame
    in load_data, we return a TimeSeriesDataset with the dataframe and the args from the init
• changed required __ init __.py files

Fixes #528 #529

Before submitting

• Was this discussed/approved via a Github issue? (no need for typos and docs improvements)
• Did you read the contributor guideline, Pull Request section?
• Did you make sure your PR does only one thing, instead of bundling different changes together?
• Did you make sure to update the documentation with your changes?
• Did you write any new necessary tests? [not needed for typos/docs]
• Did you verify new and existing tests pass locally with your changes?
• If you made a notable change (that affects users), did you update the CHANGELOG?

PR review

• Is this pull request ready for review? (if not, please submit in draft mode)

Anyone in the community is free to review the PR once the tests have passed.
If we didn't discuss your PR in Github issues there's a high chance it will not be merged.

Did you have fun?

Make sure you had fun coding 🙃

[codecov]

Codecov Report

Merging #588 (b147278) into master (27cc06d) will decrease coverage by 12.35%.
The diff coverage is 71.42%.

@@             Coverage Diff             @@
##           master     #588       +/-   ##
===========================================
- Coverage   91.41%   79.06%   -12.36%     
===========================================
  Files         117      132       +15     
  Lines        7200     7829      +629     
===========================================
- Hits         6582     6190      -392     
- Misses        618     1639     +1021     

Flag	Coverage Δ
unittests	79.06% <71.42%> (-12.36%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
flash/tabular/data.py	93.63% <ø> (ø)
flash/tabular/regression/data.py	66.66% <64.00%> (-33.34%)	⬇️
flash/tabular/classification/data.py	80.00% <76.92%> (-20.00%)	⬇️
flash/tabular/__init__.py	100.00% <100.00%> (ø)
flash/tabular/classification/__init__.py	100.00% <100.00%> (ø)
flash/tabular/regression/__init__.py	100.00% <100.00%> (ø)
flash/image/detection/data.py	29.77% <0.00%> (-65.58%)	⬇️
flash/image/detection/serialization.py	26.92% <0.00%> (-65.53%)	⬇️
flash/core/serve/_compat/cached_property.py	45.23% <0.00%> (-54.77%)	⬇️
flash/image/embedding/model.py	42.10% <0.00%> (-40.36%)	⬇️
... and 52 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 27cc06d...b147278. Read the comment docs.

[tchaton]

Should check if pytorch_forecasting is available.

[tchaton]

I find it confusing that TabularRegression is used to timeseries.
This should be a TimeSeriesForecasting Task using from_data_frame or from_csv function :)

[ethanwharris]

Hi @sumanmichael great work so far! It's definitely getting there. I think it would be simpler to just extend the base Preprocess and DataSource classes for tabular regression since regression and classification don't share many arguments or any functionality.

[ethanwharris]

Maybe this would be better just extending DataSource as it doesn't use any of the functionality from the TabularDataFrameDataSource

[ethanwharris]

Same here, maybe should just be a Preprocess? That way you can remove the arguments to init that are not used for regression

[ethanwharris]

I think this should be super().from_data_source("data_frame", ...

[tchaton]

Maybe this structure would be more appropriate:

tabular
    | _ forecasting
            | _ model.py: TabularForecaster

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.