annealing / annealing.py
mstz's picture
Update annealing.py
f10f46d
raw
history blame
6.45 kB
from typing import List
import datasets
import pandas
VERSION = datasets.Version("1.0.0")
_BASE_FEATURE_NAMES = [
"family",
"product_type",
"steel_type",
"carbon",
"hardness",
"temper_rolling",
"condition",
"formability",
"strength",
"non_ageing",
"surface_finish",
"surface_quality",
"enamelability",
"bc",
"bf",
"bt",
"bw_time",
"bl",
"m",
"chrom",
"phos",
"cbond",
"marvi",
"exptl",
"ferro",
"corr",
"blue",
"lustre",
"jurofm",
"s",
"p",
"is_coil",
"thick",
"width",
"len",
"oil",
"bore",
"packing",
"class"
]
DESCRIPTION = "Anneal dataset from the UCI ML repository."
_HOMEPAGE = "https://archive-beta.ics.uci.edu/dataset/3/annealing"
_URLS = ("https://archive-beta.ics.uci.edu/dataset/3/annealing")
_CITATION = """"""
# Dataset info
urls_per_split = {
"train": "https://huggingface.co/datasets/mstz/anneal/raw/main/anneal.data",
"test": "https://huggingface.co/datasets/mstz/anneal/raw/main/anneal.test"
}
features_types_per_config = {
"annealing": {
"family": datasets.Value("string"),
"steel_type": datasets.Value("string"),
"carbon": datasets.Value("float64"),
"hardness": datasets.Value("float64"),
"temper_rolling": datasets.Value("bool"),
"condition": datasets.Value("string"),
"formability": datasets.Value("int8"),
"strength": datasets.Value("float64"),
"surface_finish": datasets.Value("string"),
"surface_quality": datasets.Value("string"),
"enamelability": datasets.Value("int8"),
"bc": datasets.Value("bool"),
"bf": datasets.Value("bool"),
"bt": datasets.Value("bool"),
"bw_time": datasets.Value("string"),
"bl": datasets.Value("bool"),
"chrom": datasets.Value("bool"),
"phos": datasets.Value("bool"),
"cbond": datasets.Value("bool"),
"marvi": datasets.Value("bool"),
"exptl": datasets.Value("bool"),
"ferro": datasets.Value("bool"),
"corr": datasets.Value("bool"),
"blue": datasets.Value("string"),
"lustre": datasets.Value("bool"),
"jurofm": datasets.Value("bool"),
"s": datasets.Value("bool"),
"p": datasets.Value("bool"),
"is_coil": datasets.Value("bool"),
"thick": datasets.Value("float64"),
"width": datasets.Value("float64"),
"len": datasets.Value("float64"),
"oil": datasets.Value("string"),
"bore": datasets.Value("string"),
"packing": datasets.Value("string"),
"class": datasets.ClassLabel(num_classes=6, names=["1", "2", "3", "4", "5", "U"])
}
}
features_per_config = {k: datasets.Features(features_types_per_config[k]) for k in features_types_per_config}
class AnnealConfig(datasets.BuilderConfig):
def __init__(self, **kwargs):
super(AnnealConfig, self).__init__(version=VERSION, **kwargs)
self.features = features_per_config[kwargs["name"]]
class Anneal(datasets.GeneratorBasedBuilder):
# dataset versions
DEFAULT_CONFIG = "annealing"
BUILDER_CONFIGS = [
AnnealConfig(name="annealing",
description="Anneal for multiclass 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"]}),
datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs={"filepath": downloads["test"]})
]
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.columns = _BASE_FEATURE_NAMES
# data = data[data.formability != "?"]
# data = data[data.enamelability != "?"]
data.drop("product_type", axis="columns", inplace=True)
data.drop("non_ageing", axis="columns", inplace=True)
data.drop("m", axis="columns", inplace=True)
data.loc[:, "formability"] = data.formability.apply(lambda x: int(x) if x not in ("?", "-") else 0)
data.loc[:, "enamelability"] = data.enamelability.apply(lambda x: int(x) if x not in ("?", "-") else 0)
# data.loc[:, "non_ageing"] = data.formability.apply(lambda x: True if x == "-" else False)
data.loc[:, "bc"] = data.bc.apply(lambda x: True if x == "Y" else False)
data.loc[:, "bf"] = data.bf.apply(lambda x: True if x == "Y" else False)
data.loc[:, "bt"] = data.bt.apply(lambda x: True if x == "Y" else False)
data.loc[:, "bl"] = data.bl.apply(lambda x: True if x == "Y" else False)
# data.loc[:, "m"] = data.m.apply(lambda x: True if x == "Y" else False)
data.loc[:, "chrom"] = data.chrom.apply(lambda x: True if x == "C" else False)
data.loc[:, "phos"] = data.phos.apply(lambda x: True if x == "P" else False)
data.loc[:, "cbond"] = data.cbond.apply(lambda x: True if x == "Y" else False)
data.loc[:, "marvi"] = data.marvi.apply(lambda x: True if x == "Y" else False)
data.loc[:, "exptl"] = data.exptl.apply(lambda x: True if x == "Y" else False)
data.loc[:, "ferro"] = data.ferro.apply(lambda x: True if x == "Y" else False)
data.loc[:, "corr"] = data["corr"].apply(lambda x: True if x == "Y" else False)
data.loc[:, "lustre"] = data.lustre.apply(lambda x: True if x == "Y" else False)
data.loc[:, "jurofm"] = data.jurofm.apply(lambda x: True if x == "Y" else False)
data.loc[:, "s"] = data.s.apply(lambda x: True if x == "Y" else False)
data.loc[:, "p"] = data.p.apply(lambda x: True if x == "Y" else False)
data.loc[:, "class"] = data.p.apply(lambda x: int(x) if x != "U" else 0)
return data