ewstools 0.0.1

Compute early warning signals from time-series data

ewstools

A module to compute early warning signals (EWS) from time-series data. Dependencies include:

• standard python libraries: numpy, pandas, scipy, matplotlib
• LMFIT: download here

ews_compute.py

File for function ews_compute.
ews_compute takes in Series data and outputs user-specified EWS in a DataFrame.

Input (default value)

• raw_series : pandas Series indexed by time
• roll_window (0.25) : size of the rolling window (as a proportion of the length of the data)
• smooth (True) : if True, series data is detrended with a Gaussian kernel
• band_width (0.2) : bandwidth of Gaussian kernel
• ews ( ['var', 'ac', 'smax'] ) : list of strings corresponding to the desired EWS. Options include
  • 'var' : Variance
  • 'ac' : Autocorrelation
  • 'sd' : Standard deviation
  • 'cv' : Coefficient of variation
  • 'skew' : Skewness
  • 'kurt' : Kurtosis
  • 'smax' : Peak in the power spectrum
  • 'cf' : Coherence factor
  • 'aic' : AIC weights
• lag_times ( [1] ) : list of integers corresponding to the desired lag times for AC
• ham_length (40) : length of the Hamming window (used to compute power spectrum)
• ham_offset (0.5) : offset of Hamming windows as a proportion of ham_length
• w_cutoff (1) : cutoff frequency (as a proportion of maximum frequency attainable from data)

Output

• DataFrame indexed by time with columns corresponding to each EWS

ews_compute_run.py

An example script that runs ews_compute on times-series data of a stochastic simulation of May's harvesting model. It also shows how to compute kendall tau values and plot results. This can be used as a template for EWS of times-series data.

ews_compute_runMulti.py

An example script that runs ews_compute on multiple time-series data and outputs EWS as a distribution over realisations.