Transformer

BKFilter

Filter a times series using the Baxter-King filter.

This is a wrapper around the bkfilter function from statsmodels. (see statsmodels.tsa.filters.bk_filter.bkfilter).

The Baxter-King filter is intended for economic and econometric time series data and deals with the periodicity of the business cycle. Applying their band-pass filter to a series will produce a new series that does not contain fluctuations at a higher or lower frequency than those of the business cycle. Baxter-King follow Burns and Mitchell’s work on business cycles, which suggests that U.S. business cycles typically last from 1.5 to 8 years.

Schnellstart

python

from sktime.transformations.bkfilter import BKFilter

estimator = BKFilter(low=6, high=32, K=12)

Parameter(3)

lowfloat: Minimum period for oscillations. Baxter and King recommend a value of 6 for quarterly data and 1.5 for annual data.
highfloat: Maximum period for oscillations. BK recommend 32 for U.S. business cycle quarterly data and 8 for annual data.
Kint: Lead-lag length of the filter. Baxter and King suggest a truncation length of 12 for quarterly data and 3 for annual data.

Beispiele

>>> from sktime.transformations.bkfilter import BKFilter
>>> import pandas as pd
>>> import statsmodels.api as sm
>>> dta = sm. datasets. macrodata. load_pandas (). data
>>> index = pd. date_range (start = '1959Q1', end = '2009Q4', freq = 'Q')
>>> dta. set_index (index, inplace = True)
>>> bk = BKFilter (6, 24, 12)
>>> cycles = bk. fit_transform (X = dta [['realinv' ]])

Referenzen

Baxter, M. and R. G. King. “Measuring Business Cycles: Approximate: Band-Pass Filters for Economic Time Series.” Review of Economics and Statistics, 1999, 81(4), 575-593.