In [1]:

    

from scipy import signal
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt

%matplotlib inline


    

In [164]:

    

d = pd.read_csv('data/data_many.csv')
d.head()


    

    
Out[164]:






  

    

      

      
timestamps
      
Unnamed: 1
      
Unnamed: 2
      
Unnamed: 3
      
Unnamed: 4
      
Marker
    
  
  

    

      
0
      
1.488935e+09
      
-12494.062
      
31.383
      
31.871
      
35.222
      
0
    
    

      
1
      
1.488935e+09
      
-12565.521
      
18.829
      
39.432
      
23.128
      
0
    
    

      
2
      
1.488935e+09
      
-12509.839
      
21.112
      
32.225
      
33.070
      
0
    
    

      
3
      
1.488935e+09
      
-12430.331
      
46.934
      
19.734
      
49.180
      
0
    
    

      
4
      
1.488935e+09
      
-12488.146
      
27.412
      
27.025
      
23.625
      
0
    
  

In [165]:

    

X = np.array(d.ix[:, 1:5])
sig = X[:, 0]


    

In [180]:

    

freq, y = signal.welch(sig, fs=200.0)
plt.plot(freq, np.log(y))


    

In [171]:

    

b, a = signal.butter(8, (6.5/100.0, 40.0/100), btype='bandpass')
sig_f = signal.lfilter(b, a, sig)[1000:]


    

In [187]:

    

plt.plot(sig_f)


    

    
Out[187]:





[<matplotlib.lines.Line2D at 0x7fb1b389dcf8>]






    

In [188]:

    

fftfreq, fft = signal.welch(sig_f, fs=200.0)
plt.plot(fftfreq, fft)
plt.xlim(5, 40)


    

    
Out[188]:





(5, 40)






    

In [192]:

    

_ = plt.specgram(sig, Fs=200.0)


    

In [193]:

    

_ = plt.specgram(sig_f, Fs=200.0, NFFT=256, noverlap=128)


    

In [ ]:

    




    

In [ ]:

    




    

In [ ]: