Question

我正在尝试使用pandas，NumPy和SciPy过滤大量的加速度计数据。我想自定义我的频率响应，所以我发现firwin2是一个很好的过滤选项。这是我过滤数据流的代码：

def actigraph_filter(series):
    SF=25
    Nyq=SF/2
    points = np.arange(0,13,1)
    gain = [0, .05, .4, .95, 1, .9, .85, .7, .6, .45, .35, .25, 0]
    b = signal.firwin2(numtaps=170, freq=points, gain=gain, nyq=Nyq)
    return signal.filtfilt(b, [1.0], series)

我的数据以25 Hz的速度录制，正如您所看到的，我勾勒出一个（非常）粗略的频率响应，只是为了让事情变得有效。此函数在管道中运行以读取，过滤并最终为我的数据生成计数，但是当我在一段数据上运行代码时，我收到此错误：

    File "/Users/wharper/Accelerometry-Data-Processing/geneactiv_counts.py", line 45, in actigraph_filter
    return signal.filtfilt(b, [1.0], series)
  File "/Users/wharper/accelerometry/lib/python2.7/site-packages/scipy/signal/signaltools.py", line 3087, in filtfilt
    ntaps=max(len(a), len(b)))
  File "/Users/wharper/accelerometry/lib/python2.7/site-packages/scipy/signal/signaltools.py", line 3145, in _validate_pad
    ext = odd_ext(x, edge, axis=axis)
  File "/Users/wharper/accelerometry/lib/python2.7/site-packages/scipy/signal/_arraytools.py", line 102, in odd_ext
    ext = np.concatenate((2 * left_end - left_ext,
TypeError: unsupported operand type(s) for -: 'str' and 'str'

我不太确定如何继续。有没有人有任何见解？

使用SciPy过滤加速度计数据

0 个答案: