"""Letter Dataset""" from typing import List from functools import partial import string import datasets import pandas VERSION = datasets.Version("1.0.0") _ENCODING_DICS = { "letter": {letter: i for i, letter in enumerate(string.ascii_uppercase)} } DESCRIPTION = "Letter dataset." _HOMEPAGE = "https://archive-beta.ics.uci.edu/dataset/170/letter" _URLS = ("https://archive-beta.ics.uci.edu/dataset/170/letter") _CITATION = """ @misc{misc_letter_recognition_59, author = {Slate,David}, title = {{Letter Recognition}}, year = {1991}, howpublished = {UCI Machine Learning Repository}, note = {{DOI}: \\url{10.24432/C5ZP40}} } """ # Dataset info urls_per_split = { "train": "https://huggingface.co/datasets/mstz/letter/resolve/main/letter.data" } features_types_per_config = { "letter": { "x-box": datasets.Value("int64"), "y-box": datasets.Value("int64"), "width": datasets.Value("int64"), "high": datasets.Value("int64"), "onpix": datasets.Value("int64"), "x-bar": datasets.Value("int64"), "y-bar": datasets.Value("int64"), "x2bar": datasets.Value("int64"), "y2bar": datasets.Value("int64"), "xybar": datasets.Value("int64"), "x2ybr": datasets.Value("int64"), "xy2br": datasets.Value("int64"), "x-ege": datasets.Value("int64"), "xegvy": datasets.Value("int64"), "y-ege": datasets.Value("int64"), "yegvx": datasets.Value("int64"), "letter": datasets.ClassLabel(num_classes=26) } } for i, letter in enumerate(string.ascii_uppercase): features_types_per_config[letter] = { "x-box": datasets.Value("int64"), "y-box": datasets.Value("int64"), "width": datasets.Value("int64"), "high": datasets.Value("int64"), "onpix": datasets.Value("int64"), "x-bar": datasets.Value("int64"), "y-bar": datasets.Value("int64"), "x2bar": datasets.Value("int64"), "y2bar": datasets.Value("int64"), "xybar": datasets.Value("int64"), "x2ybr": datasets.Value("int64"), "xy2br": datasets.Value("int64"), "x-ege": datasets.Value("int64"), "xegvy": datasets.Value("int64"), "y-ege": datasets.Value("int64"), "yegvx": datasets.Value("int64"), "letter": datasets.ClassLabel(num_classes=2) } features_per_config = {k: datasets.Features(features_types_per_config[k]) for k in features_types_per_config} class LetterConfig(datasets.BuilderConfig): def __init__(self, **kwargs): super(LetterConfig, self).__init__(version=VERSION, **kwargs) self.features = features_per_config[kwargs["name"]] class Letter(datasets.GeneratorBasedBuilder): # dataset versions DEFAULT_CONFIG = "letter" BUILDER_CONFIGS = [ LetterConfig(name="letter", description="Letter for multiclass classification."), LetterConfig(name="A", description="Letter for binary letter A classification."), LetterConfig(name="B", description="Letter for binary letter B classification."), LetterConfig(name="C", description="Letter for binary letter C classification."), LetterConfig(name="D", description="Letter for binary letter D classification."), LetterConfig(name="E", description="Letter for binary letter E classification."), LetterConfig(name="F", description="Letter for binary letter F classification."), LetterConfig(name="G", description="Letter for binary letter G classification."), LetterConfig(name="H", description="Letter for binary letter H classification."), LetterConfig(name="I", description="Letter for binary letter I classification."), LetterConfig(name="J", description="Letter for binary letter J classification."), LetterConfig(name="K", description="Letter for binary letter K classification."), LetterConfig(name="L", description="Letter for binary letter L classification."), LetterConfig(name="M", description="Letter for binary letter M classification."), LetterConfig(name="N", description="Letter for binary letter N classification."), LetterConfig(name="O", description="Letter for binary letter O classification."), LetterConfig(name="P", description="Letter for binary letter P classification."), LetterConfig(name="Q", description="Letter for binary letter Q classification."), LetterConfig(name="R", description="Letter for binary letter R classification."), LetterConfig(name="S", description="Letter for binary letter S classification."), LetterConfig(name="T", description="Letter for binary letter T classification."), LetterConfig(name="U", description="Letter for binary letter U classification."), LetterConfig(name="V", description="Letter for binary letter V classification."), LetterConfig(name="W", description="Letter for binary letter W classification."), LetterConfig(name="X", description="Letter for binary letter X classification."), LetterConfig(name="Y", description="Letter for binary letter Y classification."), LetterConfig(name="Z", description="Letter for binary letter Z 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) data = self.preprocess(data) for row_id, row in data.iterrows(): data_row = dict(row) yield row_id, data_row def preprocess(self, data: pandas.DataFrame) -> pandas.DataFrame: for feature in _ENCODING_DICS: encoding_function = partial(self.encode, feature) data.loc[:, feature] = data[feature].apply(encoding_function) if self.config.name == "A": data.letter = data.letter.apply(lambda x: 1 if x == 0 else 0) elif self.config.name == "B": data.letter = data.letter.apply(lambda x: 1 if x == 1 else 0) elif self.config.name == "C": data.letter = data.letter.apply(lambda x: 1 if x == 2 else 0) elif self.config.name == "D": data.letter = data.letter.apply(lambda x: 1 if x == 3 else 0) elif self.config.name == "E": data.letter = data.letter.apply(lambda x: 1 if x == 4 else 0) elif self.config.name == "F": data.letter = data.letter.apply(lambda x: 1 if x == 5 else 0) elif self.config.name == "G": data.letter = data.letter.apply(lambda x: 1 if x == 6 else 0) elif self.config.name == "H": data.letter = data.letter.apply(lambda x: 1 if x == 7 else 0) elif self.config.name == "I": data.letter = data.letter.apply(lambda x: 1 if x == 8 else 0) elif self.config.name == "J": data.letter = data.letter.apply(lambda x: 1 if x == 9 else 0) elif self.config.name == "K": data.letter = data.letter.apply(lambda x: 1 if x == 10 else 0) elif self.config.name == "L": data.letter = data.letter.apply(lambda x: 1 if x == 11 else 0) elif self.config.name == "M": data.letter = data.letter.apply(lambda x: 1 if x == 12 else 0) elif self.config.name == "N": data.letter = data.letter.apply(lambda x: 1 if x == 13 else 0) elif self.config.name == "O": data.letter = data.letter.apply(lambda x: 1 if x == 14 else 0) elif self.config.name == "P": data.letter = data.letter.apply(lambda x: 1 if x == 15 else 0) elif self.config.name == "Q": data.letter = data.letter.apply(lambda x: 1 if x == 16 else 0) elif self.config.name == "R": data.letter = data.letter.apply(lambda x: 1 if x == 17 else 0) elif self.config.name == "S": data.letter = data.letter.apply(lambda x: 1 if x == 18 else 0) elif self.config.name == "T": data.letter = data.letter.apply(lambda x: 1 if x == 19 else 0) elif self.config.name == "U": data.letter = data.letter.apply(lambda x: 1 if x == 20 else 0) elif self.config.name == "V": data.letter = data.letter.apply(lambda x: 1 if x == 21 else 0) elif self.config.name == "W": data.letter = data.letter.apply(lambda x: 1 if x == 22 else 0) elif self.config.name == "X": data.letter = data.letter.apply(lambda x: 1 if x == 23 else 0) elif self.config.name == "Y": data.letter = data.letter.apply(lambda x: 1 if x == 24 else 0) elif self.config.name == "Z": data.letter = data.letter.apply(lambda x: 1 if x == 25 else 0) return data[list(features_types_per_config[self.config.name].keys())] def encode(self, feature, value): if feature in _ENCODING_DICS: return _ENCODING_DICS[feature][value] raise ValueError(f"Unknown feature: {feature}")