Logo

Using dates with timeseries models

In [1]: import statsmodels.api as sm

In [2]: import pandas

Getting started

In [3]: data = sm.datasets.sunspots.load()

Right now an annual date series must be datetimes at the end of the year.

In [4]: dates = sm.tsa.datetools.dates_from_range('1700', length=len(data.endog))

Using Pandas

Make a pandas TimeSeries or DataFrame

In [5]: endog = pandas.TimeSeries(data.endog, index=dates)

and instantiate the model

In [6]: ar_model = sm.tsa.AR(endog, freq='A')

In [7]: pandas_ar_res = ar_model.fit(maxlag=9, method='mle', disp=-1)

Let’s do some out-of-sample prediction

In [8]: pred = pandas_ar_res.predict(start='2005', end='2015')

In [9]: print pred
2005-12-31    20.003287
2006-12-31    24.703991
2007-12-31    20.026132
2008-12-31    23.473647
2009-12-31    30.858579
2010-12-31    61.335471
2011-12-31    87.024716
2012-12-31    91.321277
2013-12-31    79.921645
2014-12-31    60.799538
2015-12-31    40.374895
Freq: A-DEC, dtype: float64

Using explicit dates

In [10]: ar_model = sm.tsa.AR(data.endog, dates=dates, freq='A')

In [11]: ar_res = ar_model.fit(maxlag=9, method='mle', disp=-1)

In [12]: pred = ar_res.predict(start='2005', end='2015')

In [13]: print pred
[ 20.0033  24.704   20.0261  23.4736  30.8586  61.3355  87.0247  91.3213
  79.9216  60.7995  40.3749]

This just returns a regular array, but since the model has date information attached, you can get the prediction dates in a roundabout way.

In [14]: print ar_res.data.predict_dates
<class 'pandas.tseries.index.DatetimeIndex'>
[2005-12-31 00:00:00, ..., 2015-12-31 00:00:00]
Length: 11, Freq: A-DEC, Timezone: None

This attribute only exists if predict has been called. It holds the dates associated with the last call to predict. .. TODO: should this be attached to the results instance?

Table Of Contents

This Page