liver.py

"""ILPD"""

from typing import List

import datasets

import pandas


VERSION = datasets.Version("1.0.0")

DESCRIPTION = "ILPD dataset from the UCI ML repository."
_HOMEPAGE = "https://archive.ics.uci.edu/ml/datasets/ILPD"
_URLS = ("https://archive.ics.uci.edu/ml/datasets/ILPD")
_CITATION = """
@misc{misc_ilpd_(indian_liver_patient_dataset)_225,
  author       = {Ramana,Bendi & Venkateswarlu,N.},
  title        = {{ILPD (Indian Liver Patient Dataset)}},
  year         = {2012},
  howpublished = {UCI Machine Learning Repository},
  note         = {{DOI}: \\url{10.24432/C5D02C}}
}"""

# Dataset info
urls_per_split = {
    "train": "https://huggingface.co/datasets/mstz/liver/raw/main/Indian%20Liver%20Patient%20Dataset%20(ILPD).csv"
}
features_types_per_config = {
    "liver": {
        "age": datasets.Value("int64"),
        "is_male": datasets.Value("bool"),
        "total_bilirubin": datasets.Value("float64"),
        "direct_ribilubin": datasets.Value("float64"),
        "alkaline_phosphotase": datasets.Value("int64"),
        "alamine_aminotransferasi": datasets.Value("int64"),
        "aspartate_aminotransferase": datasets.Value("int64"),
        "total_proteins": datasets.Value("float64"),
        "albumin": datasets.Value("float64"),
        "albumin_to_globulin_ratio": datasets.Value("float64"),
        "class": datasets.ClassLabel(num_classes=2, names=("no", "yes"))
    }
}
features_per_config = {k: datasets.Features(features_types_per_config[k]) for k in features_types_per_config}


class ILPDConfig(datasets.BuilderConfig):
    def __init__(self, **kwargs):
        super(ILPDConfig, self).__init__(version=VERSION, **kwargs)
        self.features = features_per_config[kwargs["name"]]


class ILPD(datasets.GeneratorBasedBuilder):
    # dataset versions
    DEFAULT_CONFIG = "liver"
    BUILDER_CONFIGS = [
        ILPDConfig(name="liver",
                    description="ILPD for binary classification.")
    ]

    def _info(self):
        info = datasets.DatasetInfo(description=DESCRIPTION, citation=_CITATION, homepage=_HOMEPAGE,
                                    features=features_per_config[self.config.name])

        return info
    
    def _split_generators(self, dl_manager: datasets.DownloadManager) -> List[datasets.SplitGenerator]:
        downloads = dl_manager.download_and_extract(urls_per_split)

        return [
            datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepath": downloads["train"]})
        ]
    
    def _generate_examples(self, filepath: str):
        data = pandas.read_csv(filepath).infer_objects()
        data[["is_male"]].applymap(bool)
        data.loc[:, "class"] = data["class"].apply(lambda x: x - 1)
        data = data.astype({"is_male": "bool"})

        for row_id, row in data.iterrows():
            data_row = dict(row)

            yield row_id, data_row