airbnb.py

import datasets
from enum import Enum
from dataclasses import dataclass
from typing import List
import pandas as pd

logger = datasets.logging.get_logger(__name__)


_CITATION = """\
@dataset{gyodi_kristof_2021_4446043,
  author       = {Gyódi, Kristóf and
                  Nawaro, Łukasz},
  title        = {{Determinants of Airbnb prices in European cities: 
                   A spatial econometrics approach (Supplementary
                   Material)}},
  month        = jan,
  year         = 2021,
  note         = {{This research was supported by National Science 
                   Centre, Poland: Project number 2017/27/N/HS4/00951}},
  publisher    = {Zenodo},
  doi          = {10.5281/zenodo.4446043},
  url          = {https://doi.org/10.5281/zenodo.4446043}
}"""

_DESCRIPTION = """
"""

_CITIES = [
    "Amsterdam",
    "Athens",
    "Barcelona",
    "Berlin",
    "Budapest",
    "Lisbon",
    "London",
    "Paris",
    "Rome",
    "Vienna"
]

_BASE_URL = "https://zenodo.org/record/4446043/files/"
_URL_TEMPLATE = _BASE_URL + "{city}_{day_type}.csv"

class DayType(str, Enum):
    WEEKDAYS = "weekdays"
    WEEKENDS = "weekends"


@dataclass
class AirbnbFile:
    """A file from the Airbnb dataset."""

    city: str
    day_type: DayType
    @property
    def url(self) -> str:
        return _URL_TEMPLATE.format(city=self.city.lower(), day_type=self.day_type.value)



class AirbnbConfig(datasets.BuilderConfig):
    """BuilderConfig for Airbnb."""

    def __init__(self, files: List[AirbnbFile], **kwargs):
        """BuilderConfig for Airbnb.
        Args:
          **kwargs: keyword arguments forwarded to super.
        """
        super(AirbnbConfig, self).__init__(**kwargs)
        self.files = files

_WEEKDAY_FILES = [AirbnbFile(city=city, day_type=DayType.WEEKDAYS) for city in _CITIES]
_WEEKEND_FILES = [AirbnbFile(city=city, day_type=DayType.WEEKENDS) for city in _CITIES]

class Airbnb(datasets.GeneratorBasedBuilder):
    """"""

    BUILDER_CONFIGS = [
        AirbnbConfig(
            name=DayType.WEEKDAYS.value,
            files=_WEEKDAY_FILES,
        ),
        AirbnbConfig(
            name=DayType.WEEKENDS.value,
            files=_WEEKEND_FILES,
        ),
        AirbnbConfig(
            name="all",
            files=_WEEKDAY_FILES + _WEEKEND_FILES,
        ),
    ]

    def _info(self):
        features = datasets.Features(
                {
                    "_id": datasets.Value("string"),
                    "city": datasets.Value("string"),
                    "realSum": datasets.Value(dtype="float64"),
                    "room_type": datasets.Value(dtype="string"),
                    "room_shared": datasets.Value(dtype="bool"),
                    "room_private": datasets.Value(dtype="bool"),
                    "person_capacity": datasets.Value(dtype="float64"),
                    "host_is_superhost": datasets.Value(dtype="bool"),
                    "multi": datasets.Value(dtype="int64"),
                    "biz": datasets.Value(dtype="int64"),
                    "cleanliness_rating": datasets.Value(dtype="float64"),
                    "guest_satisfaction_overall": datasets.Value(dtype="float64"),
                    "bedrooms": datasets.Value(dtype="int64"),
                    "dist": datasets.Value(dtype="float64"),
                    "metro_dist": datasets.Value(dtype="float64"),
                    "attr_index": datasets.Value(dtype="float64"),
                    "attr_index_norm": datasets.Value(dtype="float64"),
                    "rest_index": datasets.Value(dtype="float64"),
                    "rest_index_norm": datasets.Value(dtype="float64"),
                    "lng": datasets.Value(dtype="float64"),
                    "lat": datasets.Value(dtype="float64")
                })
        if self.config.name == "all":
            features["day_type"] = datasets.Value(dtype="string")
        
        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=features,
            supervised_keys=None,
            homepage="https://zenodo.org/record/4446043#.ZEV8d-zMI-R",
            citation=_CITATION
        )

    def _split_generators(self, dl_manager):
        config_files: List[AirbnbFile] = self.config.files
        urls = [file.url for file in config_files]
        downloaded_files = dl_manager.download_and_extract(urls)

        return [
            datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"paths": downloaded_files})
        ]

    def _generate_examples(self, paths: List[str]):
        _id = 0
        config_files: List[AirbnbFile] = self.config.files
        include_day_type = self.config.name == "all"
        for file, path in zip(config_files, paths):
            logger.info("generating examples from = %s", path)
            df = pd.read_csv(path, index_col=0, header=0)
            for row in df.itertuples():
                city = file.city
                data = {
                    "_id": _id,
                    "city": city,
                    "realSum": row.realSum,
                    "room_type": row.room_type,
                    "room_shared": row.room_shared,
                    "room_private": row.room_private,
                    "person_capacity": row.person_capacity,
                    "host_is_superhost": row.host_is_superhost,
                    "multi": row.multi,
                    "biz": row.biz,
                    "cleanliness_rating": row.cleanliness_rating,
                    "guest_satisfaction_overall": row.guest_satisfaction_overall,
                    "bedrooms": row.bedrooms,
                    "dist": row.dist,
                    "metro_dist": row.metro_dist,
                    "attr_index": row.attr_index,
                    "attr_index_norm": row.attr_index_norm,
                    "rest_index": row.rest_index,
                    "rest_index_norm": row.rest_index_norm,
                    "lng": row.lng,
                    "lat": row.lat
                }
                if include_day_type:
                    data["day_type"] = file.day_type.value
                yield _id, data
                _id += 1