landsat.py

"""Landsat Dataset"""

from typing import List
from functools import partial

import datasets

import pandas


VERSION = datasets.Version("1.0.0")

_ENCODING_DICS = {}

DESCRIPTION = "Landsat dataset."
_HOMEPAGE = "https://archive-beta.ics.uci.edu/dataset/78/page+blocks+classification"
_URLS = ("https://archive-beta.ics.uci.edu/dataset/78/page+blocks+classification")
_CITATION = """
@misc{misc_statlog_(landsat_satellite)_146,
  author       = {Srinivasan,Ashwin},
  title        = {{Statlog (Landsat Satellite)}},
  year         = {1993},
  howpublished = {UCI Machine Learning Repository},
  note         = {{DOI}: \\url{10.24432/C55887}}
}
"""

# Dataset info
urls_per_split = {
    "train": "https://huggingface.co/datasets/mstz/landsat/raw/main/landsat.csv"
}
features_types_per_config = {
    "landsat": {
        "f1": datasets.Value("int32"),
        "f2": datasets.Value("int32"),
        "f3": datasets.Value("int32"),
        "f4": datasets.Value("int32"),
        "class": datasets.ClassLabel(num_classes=6),
    },
    "landsat_0": {
        "f1": datasets.Value("int32"),
        "f2": datasets.Value("int32"),
        "f3": datasets.Value("int32"),
        "f4": datasets.Value("int32"),
        "class": datasets.ClassLabel(num_classes=2),
    },
    "landsat_1": {
        "f1": datasets.Value("int32"),
        "f2": datasets.Value("int32"),
        "f3": datasets.Value("int32"),
        "f4": datasets.Value("int32"),
        "class": datasets.ClassLabel(num_classes=2),
    },
    "landsat_2": {
        "f1": datasets.Value("int32"),
        "f2": datasets.Value("int32"),
        "f3": datasets.Value("int32"),
        "f4": datasets.Value("int32"),
        "class": datasets.ClassLabel(num_classes=2),
    },
    "landsat_3": {
        "f1": datasets.Value("int32"),
        "f2": datasets.Value("int32"),
        "f3": datasets.Value("int32"),
        "f4": datasets.Value("int32"),
        "class": datasets.ClassLabel(num_classes=2),
    },
    "landsat_4": {
        "f1": datasets.Value("int32"),
        "f2": datasets.Value("int32"),
        "f3": datasets.Value("int32"),
        "f4": datasets.Value("int32"),
        "class": datasets.ClassLabel(num_classes=2),
    },
    "landsat_5": {
        "f1": datasets.Value("int32"),
        "f2": datasets.Value("int32"),
        "f3": datasets.Value("int32"),
        "f4": datasets.Value("int32"),
        "class": datasets.ClassLabel(num_classes=2),
    },
}
features_per_config = {k: datasets.Features(features_types_per_config[k]) for k in features_types_per_config}


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


class Landsat(datasets.GeneratorBasedBuilder):
    # dataset versions
    DEFAULT_CONFIG = "landsat"
    BUILDER_CONFIGS = [
        LandsatConfig(name="landsat", description="Landsat for multiclass classification."),
        LandsatConfig(name="landsat_0", description="Landsat for binary classification."),
        LandsatConfig(name="landsat_1", description="Landsat for binary classification."),
        LandsatConfig(name="landsat_2", description="Landsat for binary classification."),
        LandsatConfig(name="landsat_3", description="Landsat for binary classification."),
        LandsatConfig(name="landsat_4", description="Landsat for binary classification."),
        LandsatConfig(name="landsat_5", description="Landsat for 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):
        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:
        if self.config.name == "landsat_0":
            data["class"] = data["class"].apply(lambda x: 1 if x == 0 else 0)
        elif self.config.name == "landsat_1":
            data["class"] = data["class"].apply(lambda x: 1 if x == 1 else 0)
        elif self.config.name == "landsat_2":
            data["class"] = data["class"].apply(lambda x: 1 if x == 2 else 0)
        elif self.config.name == "landsat_3":
            data["class"] = data["class"].apply(lambda x: 1 if x == 3 else 0)
        elif self.config.name == "landsat_4":
            data["class"] = data["class"].apply(lambda x: 1 if x == 4 else 0)
        elif self.config.name == "landsat_5":
            data["class"] = data["class"].apply(lambda x: 1 if x == 5 else 0)

        for feature in _ENCODING_DICS:
            encoding_function = partial(self.encode, feature)
            data.loc[:, feature] = data[feature].apply(encoding_function)
                
        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}")