acute_inflammation.py

from typing import List

import datasets

import pandas


VERSION = datasets.Version("1.0.0")
_BASE_FEATURE_NAMES = [
    "temperature",
    "has_nausea",
    "has_lumbar_pain",
    "has_urine_pushing",
    "has_micturition_pains",
    "has_burnt_urethra",
    "has_inflammed_bladder",
    "has_nephritis_of_renal_pelvis",
    "has_acute_inflammation"
]

DESCRIPTION = "Acute_Inflammation dataset from the UCI ML repository."
_HOMEPAGE = "https://archive.ics.uci.edu/ml/datasets/Acute_Inflammation"
_URLS = ("https://huggingface.co/datasets/mstz/acute_inflammation/raw/main/diagnosis.csv")
_CITATION = """
@misc{misc_acute_inflammations_184,
  author       = {Czerniak,Jacek},
  title        = {{Acute Inflammations}},
  year         = {2009},
  howpublished = {UCI Machine Learning Repository},
  note         = {{DOI}: \\url{10.24432/C5V59S}}
}"""

# Dataset info
urls_per_split = {
    "train": "https://huggingface.co/datasets/mstz/acute_inflammation/raw/main/diagnosis.csv"
}
features_types_per_config = {
    "inflammation": {
        "temperature": datasets.Value("float64"),
        "has_nausea": datasets.Value("bool"),
        "has_lumbar_pain": datasets.Value("bool"),
        "has_urine_pushing": datasets.Value("bool"),
        "has_micturition_pains": datasets.Value("bool"),
        "has_burnt_urethra": datasets.Value("bool"),
        "has_inflammed_bladder": datasets.Value("bool"),
        "has_nephritis_of_renal_pelvis": datasets.Value("bool"),
        "has_acute_inflammation": datasets.ClassLabel(num_classes=2)
    },
    "nephritis": {
        "temperature": datasets.Value("float64"),
        "has_nausea": datasets.Value("bool"),
        "has_lumbar_pain": datasets.Value("bool"),
        "has_urine_pushing": datasets.Value("bool"),
        "has_micturition_pains": datasets.Value("bool"),
        "has_burnt_urethra": datasets.Value("bool"),
        "has_inflammed_bladder": datasets.Value("bool"),
        "has_acute_inflammation": datasets.Value("bool"),
        "has_nephritis_of_renal_pelvis": datasets.ClassLabel(num_classes=2)
    },
    "bladder": {
        "temperature": datasets.Value("float64"),
        "has_nausea": datasets.Value("bool"),
        "has_lumbar_pain": datasets.Value("bool"),
        "has_urine_pushing": datasets.Value("bool"),
        "has_micturition_pains": datasets.Value("bool"),
        "has_burnt_urethra": datasets.Value("bool"),
        "has_acute_inflammation": datasets.Value("bool"),
        "has_nephritis_of_renal_pelvis": datasets.Value("bool"),
        "has_inflammed_bladder": datasets.ClassLabel(num_classes=2), 
    }
}
features_per_config = {k: datasets.Features(features_types_per_config[k]) for k in features_types_per_config}


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


class Acute_Inflammation(datasets.GeneratorBasedBuilder):
    # dataset versions
    DEFAULT_CONFIG = "inflammation"
    BUILDER_CONFIGS = [
        Acute_InflammationConfig(name="inflammation",
                                 description="Binary classification of inflammation."),
        Acute_InflammationConfig(name="nephritis",
                                 description="Binary classification of nephritis."),
        Acute_InflammationConfig(name="bladder",
                                 description="Binary classification of bladder inflammation."),
    ]


    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, header=None)
        data = self.preprocess(data, config=self.config.name)

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

            yield row_id, data_row

    def preprocess(self, data: pandas.DataFrame, config: str = DEFAULT_CONFIG) -> pandas.DataFrame:
        data.columns = _BASE_FEATURE_NAMES
        boolean_features = ["has_nausea", "has_lumbar_pain", "has_urine_pushing",
                                "has_micturition_pains", "has_burnt_urethra", "has_inflammed_bladder",
                                "has_nephritis_of_renal_pelvis", "has_acute_inflammation"]
        for f in boolean_features:
            data.loc[:, f] = data[f].apply(lambda x: True if x == "yes" else False)
        
        if config == "inflammation":           
            data = data.astype({"has_acute_inflammation": int})       
        elif config == "nephritis":
            data = data.astype({"has_nephritis_of_renal_pelvis": int})
        elif config == "bladder":
            data = data.astype({"has_inflammed_bladder": int})

        data = data[list(features_types_per_config[config].keys())]       
        
        return data