|
"""Diamond Dataset""" |
|
|
|
from typing import List |
|
from functools import partial |
|
|
|
import datasets |
|
|
|
import pandas |
|
|
|
|
|
VERSION = datasets.Version("1.0.0") |
|
_BASE_FEATURE_NAMES = [ |
|
"carat", |
|
"cut", |
|
"color", |
|
"clarity", |
|
"depth", |
|
"table", |
|
"price", |
|
"observation_point_on_axis_x", |
|
"observation_point_on_axis_y", |
|
"observation_point_on_axis_z" |
|
] |
|
|
|
_ENCODING_DICS = { |
|
"cut": { |
|
"Fair": 0, |
|
"Good": 1, |
|
"Very Good": 2, |
|
"Premium": 3, |
|
"Ideal": 4 |
|
}, |
|
"clarity": { |
|
"IF": 0, |
|
"VVS1": 1, |
|
"VVS2": 2, |
|
"VS1": 3, |
|
"VS2": 4, |
|
"SI1": 5, |
|
"SI2": 6, |
|
"I1": 7 |
|
} |
|
} |
|
|
|
DESCRIPTION = "Diamond quality dataset." |
|
_HOMEPAGE = "https://www.kaggle.com/datasets/ulrikthygepedersen/diamonds" |
|
_URLS = ("https://www.kaggle.com/datasets/ulrikthygepedersen/diamonds") |
|
_CITATION = """""" |
|
|
|
|
|
urls_per_split = { |
|
"train": "https://huggingface.co/datasets/mstz/diamonds/raw/main/diamonds.csv", |
|
} |
|
features_types_per_config = { |
|
"encoding": { |
|
"feature": datasets.Value("string"), |
|
"original_value": datasets.Value("string"), |
|
"encoded_value": datasets.Value("int8"), |
|
}, |
|
|
|
"cut": { |
|
"carat": datasets.Value("float32"), |
|
"color": datasets.Value("string"), |
|
"clarity": datasets.Value("float32"), |
|
"depth": datasets.Value("float32"), |
|
"table": datasets.Value("float32"), |
|
"price": datasets.Value("float32"), |
|
"observation_point_on_axis_x": datasets.Value("float32"), |
|
"observation_point_on_axis_y": datasets.Value("float32"), |
|
"observation_point_on_axis_z": datasets.Value("float32"), |
|
"cut": datasets.ClassLabel(num_classes=5, names=("Fair", "Good", "Very Good", "Premium", "Ideal")) |
|
}, |
|
|
|
"cut_binary": { |
|
"carat": datasets.Value("float32"), |
|
"color": datasets.Value("string"), |
|
"clarity": datasets.Value("float32"), |
|
"depth": datasets.Value("float32"), |
|
"table": datasets.Value("float32"), |
|
"price": datasets.Value("float32"), |
|
"observation_point_on_axis_x": datasets.Value("float32"), |
|
"observation_point_on_axis_y": datasets.Value("float32"), |
|
"observation_point_on_axis_z": datasets.Value("float32"), |
|
"cut": 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 DiamondConfig(datasets.BuilderConfig): |
|
def __init__(self, **kwargs): |
|
super(DiamondConfig, self).__init__(version=VERSION, **kwargs) |
|
self.features = features_per_config[kwargs["name"]] |
|
|
|
|
|
class Diamond(datasets.GeneratorBasedBuilder): |
|
|
|
DEFAULT_CONFIG = "cut" |
|
BUILDER_CONFIGS = [ |
|
DiamondConfig(name="encoding", description="Encoding dictionaries for discrete features."), |
|
DiamondConfig(name="cut", description="5-ary classification, predict the cut quality of the diamond."), |
|
DiamondConfig(name="cut_binary", description="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): |
|
if self.config.name == "encoding": |
|
data = self.encoding_dics() |
|
else: |
|
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 = "cut") -> pandas.DataFrame: |
|
data["clarity"] = data.clarity.apply(lambda x: x.replace("b", "").replace("'", "")) |
|
data["cut"] = data.cut.apply(lambda x: x.replace("b", "").replace("'", "")) |
|
data["color"] = data.color.astype(str) |
|
data["color"] = data.color.apply(lambda x: x[2]).replace("\"", "") |
|
|
|
for feature in _ENCODING_DICS: |
|
encoding_function = partial(self.encode, feature) |
|
data[feature] = data[feature].apply(encoding_function) |
|
|
|
data.columns = _BASE_FEATURE_NAMES |
|
data = data.drop_duplicates(subset=["carat", "color", "clarity", "depth", "table", "price", "cut"]) |
|
|
|
if self.config.name == "cut_binary": |
|
data.cut = data.cut.apply(lambda x: 0 if x <= 2 else 1) |
|
|
|
|
|
return data[list(features_types_per_config["cut"].keys())] |
|
|
|
def encode(self, feature, value): |
|
if feature in _ENCODING_DICS: |
|
return _ENCODING_DICS[feature][value] |
|
raise ValueError(f"Unknown feature: {feature}") |
|
|
|
def encoding_dics(self): |
|
data = [pandas.DataFrame([(feature, original, encoded) for original, encoded in d.items()]) |
|
for feature, d in _ENCODING_DICS.items()] |
|
data = pandas.concat(data, axis="rows").reset_index() |
|
data.drop("index", axis="columns", inplace=True) |
|
data.columns = ["feature", "original_value", "encoded_value"] |
|
|
|
return data |
|
|
