Python 3.6-Scipy引发错误：NotImplementedError：不支持输入类型'<U32'

如何解决Python 3.6-Scipy引发错误：NotImplementedError：不支持输入类型'<U32'

我想用lfilter去噪数据。我的代码：

    data = []
    with open('Accelerometer.csv',newline='') as csvfile:
        spamreader = csv.reader(csvfile,delimiter=',',quotechar='|')
        for row in spamreader:
            data.append(row)
    
    timeInMS = []
    mx = []
    my = []
    mz = []
    for rowIndex in range(0,len(data)):
        print(rowIndex)
        timeInMS.append(data[rowIndex][1])
    
        mx.append(data[rowIndex][2])
        my.append(data[rowIndex][3])
        mz.append(data[rowIndex][4])
    # Data for plotting
    
    n = 15  # the larger n is,the smoother curve will be
    b = [1.0 / n] * n
    a = 1
    
    fMx = lfilter(b,a,mx)
    fMy = lfilter(b,my)
    fMz = lfilter(b,mz)
    
    # plotting the line 1 points
    plt.plot(timeInMS,fMx,label="x")
    
    # plotting the line 2 points
    plt.plot(timeInMS,fMy,label="y")
    
    # plotting the line 3 points
    plt.plot(timeInMS,fMz,label="z")
    
    plt.xlabel('x - axis')
    plt.ylabel('y - axis')
    plt.title('Changes on Accelerometer')
    plt.legend()
    plt.show() # display a figure.

Accelerometer.csv（1000行）：

Timestamp,Milliseconds,X,Y,Z
2020-10-05 10:19:23,11,0.02633622,0.2633622,9.933543
2020-10-05 10:19:23,18,0.037110128,0.25498247,9.938332
2020-10-05 10:19:23,29,0.034715924,0.25258827,9.939528
2020-10-05 10:19:23,37,0.25139117,9.944317
2020-10-05 10:19:23,47,0.039504327,0.2621651,9.932345
2020-10-05 10:19:23,58,0.035913024,0.25378537,9.927557
2020-10-05 10:19:23,68,9.931149
2020-10-05 10:19:23,78,0.027533319,0.25857377,87,9.94312
...
...
...

错误是：

Traceback (most recent call last):
  File "Accelerometer/main.py",line 32,in <module>
    fMx = lfilter(b,mx)
  File "Accelerometer\venv\lib\site-packages\scipy\signal\signaltools.py",line 1883,in lfilter
    raise NotImplementedError("input type '%s' not supported" % dtype)
NotImplementedError: input type '<U32' not supported

解决方法

使用熊猫pd.read_csv()读取文件确实有帮助。这将自动将数据放入方便的格式。然后，您可以直接使用mx = data['X']。

在没有大熊猫的情况下，您会遇到诸如此处的问题，在这里您会忘记跳过标题，此外，您的数据全都是字符串格式而不是数字格式。还将日期和时间从字符串格式转换为实际的日期时间格式。

import pandas as pd
import matplotlib.pyplot as plt
from scipy.signal import lfilter

data = pd.read_csv('Accelerometer.csv')

n = 15  # the larger n is,the smoother curve will be
b = [1.0 / n] * n
a = 1

fMx = lfilter(b,a,data['X'])
fMy = lfilter(b,data['Y'])
fMz = lfilter(b,data['Z'])

# plotting the line 1 points
plt.plot(data['Milliseconds'],fMx,label="x")

# plotting the line 2 points
plt.plot(data['Milliseconds'],fMy,label="y")

# plotting the line 3 points
plt.plot(data['Milliseconds'],fMz,label="z")

plt.xlabel('x - axis')
plt.ylabel('y - axis')
plt.title('Changes on Accelerometer')
plt.legend()
plt.show()  # Display a figure.