doubt.datasets.facebook_metrics

  1"""Facebook metrics data set.
  2
  3This data set is from the UCI data set archive, with the description being the original
  4description verbatim. Some feature names may have been altered, based on the
  5description.
  6"""
  7
  8import io
  9import zipfile
 10
 11import pandas as pd
 12
 13from .dataset import BASE_DATASET_DESCRIPTION, BaseDataset
 14
 15
 16class FacebookMetrics(BaseDataset):
 17    __doc__ = f"""
 18    The data is related to posts' published during the year of 2014 on the Facebook's
 19    page of a renowned cosmetics brand.
 20
 21    {BASE_DATASET_DESCRIPTION}
 22
 23    Features:
 24        page_likes(int):
 25            The total number of likes of the Facebook page at the given time.
 26        post_type (int):
 27            The type of post. Here 0 means 'Photo', 1 means 'Status', 2 means 'Link'
 28            and 3 means 'Video'
 29        post_category (int):
 30            The category of the post.
 31        post_month (int):
 32            The month the post was posted, from 1 to 12 inclusive.
 33        post_weekday (int):
 34            The day of the week the post was posted, from 1 to 7 inclusive.
 35        post_hour (int):
 36            The hour the post was posted, from 0 to 23 inclusive
 37        paid (int):
 38            Binary feature, whether the post was paid for.
 39
 40    Targets:
 41        total_reach (int):
 42            The lifetime post total reach.
 43        total_impressions (int):
 44            The lifetime post total impressions.
 45        engaged_users (int):
 46            The lifetime engaged users.
 47        post_consumers (int):
 48            The lifetime post consumers.
 49        post_consumptions (int):
 50            The lifetime post consumptions.
 51        post_impressions (int):
 52            The lifetime post impressions by people who liked the page.
 53        post_reach (int):
 54            The lifetime post reach by people who liked the page.
 55        post_engagements (int):
 56            The lifetime people who have liked the page and engaged with
 57            the post.
 58        comments (int):
 59            The number of comments.
 60        shares (int):
 61            The number of shares.
 62        total_interactions (int):
 63            The total number of interactions
 64
 65    Source:
 66        https://archive.ics.uci.edu/ml/datasets/Facebook+metrics
 67
 68    Examples:
 69        Load in the data set::
 70
 71            >>> dataset = FacebookMetrics()
 72            >>> dataset.shape
 73            (500, 18)
 74
 75        Split the data set into features and targets, as NumPy arrays::
 76
 77            >>> X, y = dataset.split()
 78            >>> X.shape, y.shape
 79            ((500, 7), (500, 11))
 80
 81        Perform a train/test split, also outputting NumPy arrays::
 82
 83            >>> train_test_split = dataset.split(test_size=0.2, random_seed=42)
 84            >>> X_train, X_test, y_train, y_test = train_test_split
 85            >>> X_train.shape, y_train.shape, X_test.shape, y_test.shape
 86            ((388, 7), (388, 11), (112, 7), (112, 11))
 87
 88        Output the underlying Pandas DataFrame::
 89
 90            >>> df = dataset.to_pandas()
 91            >>> type(df)
 92            <class 'pandas.core.frame.DataFrame'>
 93    """
 94
 95    _url = (
 96        "https://archive.ics.uci.edu/ml/machine-learning-databases/"
 97        "00368/Facebook_metrics.zip"
 98    )
 99
100    _features = range(7)
101    _targets = range(7, 18)
102
103    def _prep_data(self, data: bytes) -> pd.DataFrame:
104        """Prepare the data set.
105
106        Args:
107            data (bytes): The raw data
108
109        Returns:
110            Pandas dataframe: The prepared data
111        """
112        # Convert the bytes into a file-like object
113        buffer = io.BytesIO(data)
114
115        # Unzip the file and pull out the csv file
116        with zipfile.ZipFile(buffer, "r") as zip_file:
117            csv = zip_file.read("dataset_Facebook.csv")
118
119        # Convert the bytes into a file-like object
120        csv_file = io.BytesIO(csv)
121
122        # Read the file-like object into a dataframe
123        cols = [
124            "page_likes",
125            "post_type",
126            "post_category",
127            "post_month",
128            "post_weekday",
129            "post_hour",
130            "paid",
131            "total_reach",
132            "total_impressions",
133            "engaged_users",
134            "post_consumers",
135            "post_consumptions",
136            "post_impressions",
137            "post_reach",
138            "post_engagements",
139            "comments",
140            "shares",
141            "total_interactions",
142        ]
143        df = pd.read_csv(csv_file, sep=";", names=cols, header=0, index_col=0)
144
145        # Numericalise post type
146        post_types = list(df.post_type.unique())
147        df["post_type"] = df.post_type.map(lambda txt: post_types.index(txt))
148
149        return df