Python: Data analysis using FFT

Question

I have a data which looks like this:

YYYY-MO-DD HH-MI-SS_SSS,  ATMOSPHERIC PRESSURE (hPa) mean,   ATMOSPHERIC PRESSURE (hPa) std
2016-04-20 00:00:00,1006.0515000000001,0.029159119281803602
2016-04-20 00:01:00,1006.039666666667,0.03565211699642609
2016-04-20 00:02:00,1006.0148333333334,0.036891580347842706
2016-04-20 00:03:00,1006.0058333333335,0.03351152934243721
2016-04-20 00:04:00,1005.9714999999999,0.03155973620213212
2016-04-20 00:05:00,1005.955666666667,0.027207094455343653
.............

I'm interested in the Pressure mean which is sampled every minute. My goal is to look for periodic frequencies inside the data.

I've tried the following:

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt  
from scipy.fftpack import fft
    df3 = pd.read_csv('Pressure - Dates by Minute.csv', sep=",", skiprows=0)
    Pressure = df3['ATMOSPHERIC PRESSURE (hPa) mean']
    frate = 1/60
    Pfft = np.fft.fft(Pressure[0])
    freqs = fft.fftfreq(len(Pfft), 1/frate)

But i'm getting "tuple index out of range" errors

Any ideas on how to analyze the fft and plot the matching frequencies against the raw data?

The raw data looks like this:

Thanks!

Have you verified that there is data in Pressure? Try printing out the len(Pressure) — Tom Myddeltyn
– Tom Myddeltyn, Commented May 13, 2016 at 12:14
I am assuming, but is the last line you give the one that is giving you the errors? — Tom Myddeltyn
– Tom Myddeltyn, Commented May 13, 2016 at 12:18
Your Pressure is a Pandas Series, roughly speaking a numpy ndarray and you compute its DFT by Pfft = np.fft.fft(Pressure) (no indexing!). — I don't know if this problem is or isn't related to the issue that you showed us. — gboffi
– gboffi, Commented May 13, 2016 at 12:21

Nils Werner · Accepted Answer · 2016-05-13 12:24:53Z

3

You are retrieving only the first element of Pressure but you should do the fourier analysis on all samples. If you replace

Pfft = np.fft.fft(Pressure[0])

with

Pfft = np.fft.fft(Pressure)

it works:

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt

df3 = pd.read_csv('Pressure - Dates by Minute.csv', sep=",", skiprows=0)
Pressure = df3['ATMOSPHERIC PRESSURE (hPa) mean']
frate = 1. / 60
Pfft = np.fft.fft(Pressure)
Pfft[0] = 0  # Set huge DC component to zero, equates to Pressure = Pressure - numpy.mean(Pressure)

freqs = np.fft.fftfreq(len(Pfft), 1. / frate)
plt.plot(freqs, Pfft)
plt.show()

answered May 13, 2016 at 12:24

Nils Werner

37.2k7 gold badges85 silver badges108 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

Tom Myddeltyn · Accepted Answer · 2016-05-13 12:25:11Z

1

I'm taking a stab at this, I think that the issue is that Pressure[0] is a value and you need to pass an array to np.fft.fft() so try Pfft = np.fft.fft(Pressure)

answered May 13, 2016 at 12:25

Tom Myddeltyn

1,3751 gold badge14 silver badges28 bronze badges

Collectives™ on Stack Overflow

Python: Data analysis using FFT

2 Answers 2

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related