madelon.py

from typing import List

import datasets

import pandas


VERSION = datasets.Version("1.0.0")


DESCRIPTION = "Madelon dataset from the UCI ML repository."
_HOMEPAGE = "https://archive-beta.ics.uci.edu/dataset/3/madelon"
_URLS = ("https://archive-beta.ics.uci.edu/dataset/3/madelon")
_CITATION = """"""

# Dataset info
urls_per_split = {
    "train": "https://huggingface.co/datasets/mstz/madelon/raw/main/madelon_train.csv",
    "validation": "https://huggingface.co/datasets/mstz/madelon/raw/main/madelon_valid.csv"
}
features_types_per_config = {
    "madelon": {str(i): datasets.Value("int16") for i in range(500)}
}
features_types_per_config["madelon"]["500"] = datasets.ClassLabel(num_classes=2)

features_per_config = {k: datasets.Features(features_types_per_config[k]) for k in features_types_per_config}


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


class Madelon(datasets.GeneratorBasedBuilder):
    # dataset versions
    DEFAULT_CONFIG = "madelon"
    BUILDER_CONFIGS = [
        MadelonConfig(name="madelon",
                    description="Madelon for multiclass classification.")
    ]


    def _info(self):
        if self.config.name not in features_per_config:
            raise ValueError(f"Unknown configuration: {self.config.name}")
        
        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"]}),
            datasets.SplitGenerator(name=datasets.Split.VALIDATION, gen_kwargs={"filepath": downloads["validation"]})
        ]
    
    def _generate_examples(self, filepath: str):
            data = pandas.read_csv(filepath)
            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["500"] = data["500"].apply(lambda x: max(0, x)).astype(int)
        if "0.1" in data:
            data.drop("0.1", axis="columns", inplace=True)

        return data