Carefree-data

Release Notes

`carefree-data` now uses `black` and `mypy` to format codes.

`TabularData`
+ Supported `simplify = True`, which will simplify the whole process / structures of `TabularData`.
+ Supported 'fusing' categorical features when there are too many different values (e.g. 10^8).
+ Optimized serializing. Now `carefree-data` will save only one copy of `numpy` data, instead of many duplications.

Release Notes

`TabularData`
🎉🎉 `TabularData` now supports time series datasets! 🎉🎉

> For detailed examples, please refers to time series examples in [`carefree-learn`](https://github.com/carefree0910/carefree-learn/tree/dev/examples/time_series)!

Miscellaneous fixes and updates.

Release Notes

`TabularData`

+ `skip_first` is replaced by `has_column_names`, so `column_names` will now be extracted from `csv` files automatically.
+ `quote_char` is introduced. For instance, `"` is the default `quote_char` for `csv` files.

Here's an example:

`quote_test.csv`:
text
f1,f2,f3,f4,f5
1,"2, 3",4","5,6


Here's the unittest:

python
data = TabularData().read("quote_test.csv")
self.assertDictEqual(data.column_names, {0: "f1", 1: "f2", 2: "f3", 3: "f4", 4: "f5"})
self.assertListEqual(data.raw.x[0], ["1", '"2, 3"', '4"', '"5'])
self.assertListEqual(data.raw.y[0], ["6"])


`read_file`
`read_file` now supports `contains_labels` argument

Breaking changes

The meaning of `label_idx` has changed. Previously, `None` means that current data file does not contain label column. But now, we introduced `contains_labels` as substitute, and `None` means that we don't know the exact `label_idx`, and has to infer it from other information.

Miscellaneous fixes and updates.

Miscellaneous fixes and updates.