Naming is really confusing

[Gautzilla]

🐳 SPRING CLEANING TIME

Before releasing a v1.0.0 of OSEkit, I'd like to make some changes in the API naming because lots of it seem quite confusing.
e.g. the single Dataset word means several different things, Analysis is unclear, analysis_dataset makes me wake up in tears in the middle of the night.

I'll use this issue to suggest some renamings, with a full example using a new naming convention.

If OSEkit users (@PaulCarvaillo @PCarvaillo @MaelleTtrt @mathieudpnt @ElodieENSTA @cazaudo @naiscd ...) have some feedback on these suggestions or want to add some, it'd be great!

🐬 Suggested changes

old name	new name	example
Dataset	Project	project = Project(folder=..., ...)
Analysis	Transform	transform = Transform()
AnalysisType	OutputType	transform = Transform(outputs=OutputType.AUDIO)
AnalysisType.MATRIX	OutputType.SPECTRUM
Dataset.run_analysis()	Project.run()
Dataset.get_analysis_audiodataset()	Project.prepare_audio()
Dataset.get_analysis_spectrodataset()	Project.prepare_spectrum()
core_api	core	from osekit.core.audio_dataset import AudioDataset
public_api	public	from osekit.public.transform import Transform

🐬 Exemple

It would lead to something like that, which IMO is a bit cleaner than the current semantics of the API:

from osekit.public.project import Project
from pathlib import Path
from osekit.core.audio_dataset import AudioDataset
from osekit.public.transform import Transform, OutputType

# PROJECT BUILD
project = Project(
    folder = Path(r"cool_folder"),
    ...
)
project.build()


# TRANSFORM DECLARATION
transform = Transform(
    outputs = OutputType.AUDIO | OutputType.SPECTRUM,
    ...
)

## WITH CORE FILTERING
ads = project.prepare_audio(transform=transform)
ads.data = ads.data[:10]


# TRANSFORM RUN
project.run(transform=transform, audio_dataset=ads)

[ElodieENSTA]

I'm not sure about Transform, maybe other users may have a better feeling on this.

Otherwise the other names sound good to me :)

[naiscd]

I agree, for me Analysis makes more sense than Transform, but I agree that prepare_audio() and prepare_spectrum() is more explicit !

[mathieudpnt]

I like Transform, i think you have to see it as a noun instead of a verb, it makes sense to me this way. Otherwise i would call it something like Analysis_config but let see what other user think

[Gautzilla]

Transform seems to be quite commonly used in packages such as pytorch or scikit-learn, but that doesn't mean we have to go with this!

I feel like analysis doesn't reflect the processing aspect (it sounds more like reading the audio and outputting some stats or whatever). Maybe something like processing or computation? But I think transform is a bit more direct