92 lines
3.2 KiB
Python
92 lines
3.2 KiB
Python
# -*- coding: utf-8 -*-
|
|
|
|
import numpy as np
|
|
import pandas as pd
|
|
|
|
from ..compat import DTYPE
|
|
|
|
__all__ = [
|
|
'load_austres'
|
|
]
|
|
|
|
|
|
def load_austres(as_series=False, dtype=DTYPE):
|
|
"""Quarterly residential data.
|
|
|
|
Numbers (in thousands) of Australian residents measured quarterly from
|
|
March 1971 to March 1994.
|
|
|
|
Parameters
|
|
----------
|
|
as_series : bool, optional (default=False)
|
|
Whether to return a Pandas series. If False, will return a 1d
|
|
numpy array.
|
|
|
|
dtype : type, optional (default=np.float64)
|
|
The type to return for the array. Default is np.float64, which is used
|
|
throughout the package as the default type.
|
|
|
|
Returns
|
|
-------
|
|
rslt : array-like, shape=(n_samples,)
|
|
The austres vector.
|
|
|
|
Examples
|
|
--------
|
|
>>> from pmdarima.datasets import load_austres
|
|
>>> load_austres()
|
|
np.array([13067.3, 13130.5, 13198.4, 13254.2, 13303.7, 13353.9,
|
|
13409.3, 13459.2, 13504.5, 13552.6, 13614.3, 13669.5,
|
|
13722.6, 13772.1, 13832.0, 13862.6, 13893.0, 13926.8,
|
|
13968.9, 14004.7, 14033.1, 14066.0, 14110.1, 14155.6,
|
|
14192.2, 14231.7, 14281.5, 14330.3, 14359.3, 14396.6,
|
|
14430.8, 14478.4, 14515.7, 14554.9, 14602.5, 14646.4,
|
|
14695.4, 14746.6, 14807.4, 14874.4, 14923.3, 14988.7,
|
|
15054.1, 15121.7, 15184.2, 15239.3, 15288.9, 15346.2,
|
|
15393.5, 15439.0, 15483.5, 15531.5, 15579.4, 15628.5,
|
|
15677.3, 15736.7, 15788.3, 15839.7, 15900.6, 15961.5,
|
|
16018.3, 16076.9, 16139.0, 16203.0, 16263.3, 16327.9,
|
|
16398.9, 16478.3, 16538.2, 16621.6, 16697.0, 16777.2,
|
|
16833.1, 16891.6, 16956.8, 17026.3, 17085.4, 17106.9,
|
|
17169.4, 17239.4, 17292.0, 17354.2, 17414.2, 17447.3,
|
|
17482.6, 17526.0, 17568.7, 17627.1, 17661.5])
|
|
|
|
>>> load_austres(True).head()
|
|
0 13067.3
|
|
1 13130.5
|
|
2 13198.4
|
|
3 13254.2
|
|
4 13303.7
|
|
dtype: float64
|
|
|
|
Notes
|
|
-----
|
|
This is quarterly data, so *m* should be set to 4 when using in a seasonal
|
|
context.
|
|
|
|
References
|
|
----------
|
|
.. [1] P. J. Brockwell and R. A. Davis (1996)
|
|
"Introduction to Time Series and Forecasting." Springer
|
|
"""
|
|
rslt = np.array([
|
|
13067.3, 13130.5, 13198.4, 13254.2, 13303.7, 13353.9,
|
|
13409.3, 13459.2, 13504.5, 13552.6, 13614.3, 13669.5,
|
|
13722.6, 13772.1, 13832.0, 13862.6, 13893.0, 13926.8,
|
|
13968.9, 14004.7, 14033.1, 14066.0, 14110.1, 14155.6,
|
|
14192.2, 14231.7, 14281.5, 14330.3, 14359.3, 14396.6,
|
|
14430.8, 14478.4, 14515.7, 14554.9, 14602.5, 14646.4,
|
|
14695.4, 14746.6, 14807.4, 14874.4, 14923.3, 14988.7,
|
|
15054.1, 15121.7, 15184.2, 15239.3, 15288.9, 15346.2,
|
|
15393.5, 15439.0, 15483.5, 15531.5, 15579.4, 15628.5,
|
|
15677.3, 15736.7, 15788.3, 15839.7, 15900.6, 15961.5,
|
|
16018.3, 16076.9, 16139.0, 16203.0, 16263.3, 16327.9,
|
|
16398.9, 16478.3, 16538.2, 16621.6, 16697.0, 16777.2,
|
|
16833.1, 16891.6, 16956.8, 17026.3, 17085.4, 17106.9,
|
|
17169.4, 17239.4, 17292.0, 17354.2, 17414.2, 17447.3,
|
|
17482.6, 17526.0, 17568.7, 17627.1, 17661.5]).astype(dtype)
|
|
|
|
if as_series:
|
|
return pd.Series(rslt)
|
|
return rslt
|