doubt.datasets.parkinsons

  1"""Parkinsons 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
  9
 10import pandas as pd
 11
 12from .dataset import BASE_DATASET_DESCRIPTION, BaseDataset
 13
 14
 15class Parkinsons(BaseDataset):
 16    __doc__ = f"""
 17    This dataset is composed of a range of biomedical voice measurements from 42 people
 18    with early-stage Parkinson's disease recruited to a six-month trial of a
 19    telemonitoring device for remote symptom progression monitoring. The recordings
 20    were automatically captured in the patient's homes.
 21
 22    Columns in the table contain subject number, subject age, subject gender, time
 23    interval from baseline recruitment date, motor UPDRS, total UPDRS, and 16
 24    biomedical voice measures. Each row corresponds to one of 5,875 voice recording
 25    from these individuals. The main aim of the data is to predict the motor and total
 26    UPDRS scores ('motor_UPDRS' and 'total_UPDRS') from the 16 voice measures.
 27
 28    {BASE_DATASET_DESCRIPTION}
 29
 30    Features:
 31        subject# (int):
 32            Integer that uniquely identifies each subject
 33        age (int):
 34            Subject age
 35        sex (int):
 36            Binary feature. Subject sex, with 0 being male and 1 female
 37        test_time (float):
 38            Time since recruitment into the trial. The integer part is the
 39            number of days since recruitment
 40        Jitter(%) (float):
 41            Measure of variation in fundamental frequency
 42        Jitter(Abs) (float):
 43            Measure of variation in fundamental frequency
 44        Jitter:RAP (float):
 45            Measure of variation in fundamental frequency
 46        Jitter:PPQ5 (float):
 47            Measure of variation in fundamental frequency
 48        Jitter:DDP (float):
 49            Measure of variation in fundamental frequency
 50        Shimmer (float):
 51            Measure of variation in amplitude
 52        Shimmer(dB) (float):
 53            Measure of variation in amplitude
 54        Shimmer:APQ3 (float):
 55            Measure of variation in amplitude
 56        Shimmer:APQ5 (float):
 57            Measure of variation in amplitude
 58        Shimmer:APQ11 (float):
 59            Measure of variation in amplitude
 60        Shimmer:DDA (float):
 61            Measure of variation in amplitude
 62        NHR (float):
 63            Measure of ratio of noise to tonal components in the voice
 64        HNR (float):
 65            Measure of ratio of noise to tonal components in the voice
 66        RPDE (float):
 67            A nonlinear dynamical complexity measure
 68        DFA (float):
 69            Signal fractal scaling exponent
 70        PPE (float):
 71            A nonlinear measure of fundamental frequency variation
 72
 73    Targets:
 74        motor_UPDRS (float):
 75            Clinician's motor UPDRS score, linearly interpolated
 76        total_UPDRS (float):
 77            Clinician's total UPDRS score, linearly interpolated
 78
 79    Source:
 80        https://archive.ics.uci.edu/ml/datasets/Parkinsons+Telemonitoring
 81
 82    Examples:
 83        Load in the data set::
 84
 85            >>> dataset = Parkinsons()
 86            >>> dataset.shape
 87            (5875, 22)
 88
 89        Split the data set into features and targets, as NumPy arrays::
 90
 91            >>> X, y = dataset.split()
 92            >>> X.shape, y.shape
 93            ((5875, 20), (5875, 2))
 94
 95        Perform a train/test split, also outputting NumPy arrays::
 96
 97            >>> train_test_split = dataset.split(test_size=0.2, random_seed=42)
 98            >>> X_train, X_test, y_train, y_test = train_test_split
 99            >>> X_train.shape, y_train.shape, X_test.shape, y_test.shape
100            ((4659, 20), (4659, 2), (1216, 20), (1216, 2))
101
102        Output the underlying Pandas DataFrame::
103
104            >>> df = dataset.to_pandas()
105            >>> type(df)
106            <class 'pandas.core.frame.DataFrame'>
107    """
108
109    _url = (
110        "https://archive.ics.uci.edu/ml/machine-learning-databases/"
111        "parkinsons/telemonitoring/parkinsons_updrs.data"
112    )
113
114    _features = range(20)
115    _targets = [20, 21]
116
117    def _prep_data(self, data: bytes) -> pd.DataFrame:
118        """Prepare the data set.
119
120        Args:
121            data (bytes): The raw data
122
123        Returns:
124            Pandas dataframe: The prepared data
125        """
126        # Convert the bytes into a file-like object
127        csv_file = io.BytesIO(data)
128
129        # Load in dataframe
130        df = pd.read_csv(csv_file, header=0)
131
132        # Put target columns at the end
133        cols = [col for col in df.columns if col[-5:] != "UPDRS"]
134        df = df[cols + ["motor_UPDRS", "total_UPDRS"]]
135
136        return df