Alpha-Beta Filter

The signal samples of a signal trace can be processed with an alpha-beta filter.

Create the Trace Collection

You can smooth the signal samples with an alpha-beta filter by calling the method alpha_beta_filter().

An alpha-beta filter trace collection is returned.

>>> # filter the signal samples
>>> traces = Trace('Signal', [1, 2, 3]).alpha_beta_filter(dt=1, alpha=0.5, beta=1)
>>> traces
AlphaBetaFilterTraces(forecast=Trace(label='Signal:filter:forecast',
                      samples=[1.0, 1.0, 2.5]),
  forecast_sign=Trace(label='Signal:filter:forecast_sign',
                      samples=[1.0, 1.0, 1.0]),
  level=Trace(label='Signal:filter:level',
              samples=[1.0, 1.5, 2.75]),
  level_sign=Trace(label='Signal:filter:level_sign',
                   samples=[1.0, 1.0, 1.0]),
  trend=Trace(label='Signal:filter:trend',
              samples=[0.0, 1.0, 1.5]),
  trend_sign=Trace(label='Signal:filter:trend_sign',
                   samples=[0.0, 1.0, 1.0]),
  trend_inflection=Trace(label='Signal:filter:trend_inflection',
                         samples=[0.0, 1.0, 1.0]),
  error=Trace(label='Signal:filter:error',
              samples=[0.0, 1.0, 0.5]),
  variance_forecast=Trace(label='Signal:filter:variance_forecast',
                          samples=[3.0, 3.0, 3.0]),
  variance_level=Trace(label='Signal:filter:variance_level',
                       samples=[1.0, 1.0, 1.0]),
  variance_trend=Trace(label='Signal:filter:variance_trend',
                       samples=[2.0, 2.0, 2.0]))

Note

The alpha-beta filter uses the first signal sample as the initial level, and the initial trend is set by default to zero.

Figure

You can visualize the trace collection with a figure containing the subplots of the traces in the collection by calling the method figure().

>>> figure = Trace('Signal', range(20)).alpha_beta_filter(1, 0.5, 1).figure()

(Source code, html)