Processor - Record the adjusted dtype in Metadata for datetime columns to correct the Synthesizer error

[matheme-justyn]

for #466

• feat

  • PETsARD/util/safe_dtype.py, init.py
    • create safe_infer_dtype - e0f5425
  • PETsARD/processor/base.py
    • add _adjust_metadata() in transform() - 6969af1
  • PETsARD/loader/metadata.py
    • take infer_dtype_after_preproc in to_sdv() - 9086e34

• test

  • demo/dev/Issue466.ipynb
    • create Issue466.ipynb - fb6ad1b
    • upload Issue466 for evidence - 5a146aa
    • delete Issue466 - 5d1e0db

---

Solution evidence

#465

from PETsARD import Loader, Processor, Synthesizer


for benchmark in ['bike-sales', 'energydata_complete', 'olist']:
    load = Loader(filepath=f'benchmark://{benchmark}')
    load.load()

    proc = Processor(metadata=load.metadata)
    proc.fit(data=load.data)
    preproc_data = proc.transform(data=load.data)

    syn = Synthesizer('default')
    syn.create(data=preproc_data, metadata=proc._metadata)
    syn.fit_sample()

---

#466

from PETsARD import Loader, Processor, Synthesizer


for benchmark in [
        'bike-sales',
        'energydata_complete',
        'olist'
]:
    load = Loader(filepath=f'benchmark://{benchmark}')
    load.load()

    proc = Processor(metadata=load.metadata)
    proc.fit(data=load.data)
    preproc_data = proc.transform(data=load.data)

    syn = Synthesizer('smartnoise-aim')
    syn.create(data=preproc_data, metadata=proc._metadata)
    syn.fit_sample()

---

• To see the specific tasks where the Asana app for GitHub is being used, see below:
  • https://app.asana.com/0/0/1207043084301540
  • https://app.asana.com/0/0/1207043084301548