electricity.py

import datasets
import json
import logging
import os
import pandas as pd
from typing import List

logger = logging.getLogger(__name__)


_CITATION = "TBD"

_DESCRIPTION = "TBD"

_HOMEPAGE = "TBD"

_LICENSE = "TBD"

_URL = "https://raw.githubusercontent.com/arthuractivemodeling/sandbox-data/main/electricity/data/"

_URLS = {
    "train": _URL + "electricity_train.csv",
    "val": _URL + "electricity_val.csv"
}


class Electricity(datasets.GeneratorBasedBuilder):

    VERSION = datasets.Version("0.1.0")

    BUILDER_CONFIGS = [
        datasets.BuilderConfig(name="default", version=VERSION, description="TBD")
    ]

    DEFAULT_CONFIG_NAME = "default"

    # write out the features in the order they appear in the dataframe
    FEATURES = ["date", "day", "period", "nswprice", "nswdemand", "vicprice", "vicdemand", "transfer"]
    GROUND_TRUTH = ["price_increase"]

    # separate floats from ints
    FLOAT_INPUTS = ["date", "period", "nswprice", "nswdemand", "vicprice", "vicdemand", "transfer"]
    INT_INPUTS = ["day"]

    def _info(self):
        features = datasets.Features(
            {
                **{f: datasets.Value("float64") for f in self.FLOAT_INPUTS},
                **{f: datasets.Value("int64") for f in self.INT_INPUTS},
                **{f: datasets.Value("int64") for f in self.GROUND_TRUTH}
            }
        )
        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=features,
            homepage=_HOMEPAGE,
            license=_LICENSE,
            citation=_CITATION,
        )

    def _split_generators(self, dl_manager: datasets.DownloadManager) -> List[datasets.SplitGenerator]:
        downloaded_files = dl_manager.download_and_extract(_URLS)

        return [
            datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepath": "electricity_train.csv"}),
            datasets.SplitGenerator(name=datasets.Split.VALIDATION, gen_kwargs={"filepath": "electricity_val.csv"}),
        ]

    def _generate_examples(self, filepath):
        data = pd.read_csv(_URL + filepath, index_col=0)
        for row_id, row in data.iterrows():
            yield row_id, dict(row)