https://github.com/ariaghora/robust-period

Unofficial Implementation of RobustPeriod: Time-Frequency Mining for Robust Multiple Periodicities Detection
https://github.com/ariaghora/robust-period

Last synced: about 2 months ago
JSON representation

Unofficial Implementation of RobustPeriod: Time-Frequency Mining for Robust Multiple Periodicities Detection

• Host: GitHub
• URL: https://github.com/ariaghora/robust-period
• Owner: ariaghora
• Created: 2021-04-29T16:59:57.000Z (over 3 years ago)
• Default Branch: master
• Last Pushed: 2023-03-26T13:53:06.000Z (over 1 year ago)
• Last Synced: 2024-06-01T23:35:06.463Z (4 months ago)
• Language: Python
• Size: 1.08 MB
• Stars: 88
• Watchers: 2
• Forks: 13
• Open Issues: 2
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
Unofficial Implementation of RobustPeriod: Time-Frequency Mining for Robust Multiple Periodicities Detection. 

> Please note that I could not fully replicate the paper, especially the optimization part. 

    In fact, I used a different optimization method rather than the said ADMM (the timing result will be greatly different from ones presented in the paper). 

    It is either the details in paper are insufficient or my understanding was lacking. I welcome any contribution. 

## Installation

`pip install --upgrade git+https://github.com/ariaghora/robust-period.git`

## Usage example

```python

import numpy as np

import matplotlib.pyplot as plt

from robustperiod import robust_period, robust_period_full, plot_robust_period

from robustperiod.utils import sinewave, triangle

from statsmodels.datasets.co2.data import load_pandas

m = 1000

y1 = sinewave(m, 20, 1)

y2 = sinewave(m, 50, 1)

y3 = sinewave(m, 100, 1)

tri = triangle(m, 10)

noise = np.random.normal(0, 0.1, m)

y = y1+y2+y3+tri+noise

y[m // 2] += 10  # sudden spike

plt.plot(y)

plt.title('Dummy dataset')

plt.show()

lmb = 1e+6

c = 2

num_wavelets = 8

zeta = 1.345

periods, W, bivar, periodograms, p_vals, ACF = robust_period_full(

    y, 'db10', num_wavelets, lmb, c, zeta)

plot_robust_period(periods, W, bivar, periodograms, p_vals, ACF)

```

## Input



  



## Output



  





  



Please note that I hacked some parts of result presentation code so the results match the paper as close as possible.