Shape distribution

This function is used for distribution analysis of any type. It generates return the shape and scale of a distribution. Inputs can be:

-only magnitude (omni-directional extreme value ananlysis) -magnitute and direction (directional ARI with omni or directional POT)_

magstr

Name of the column from which to get stats.

drrstr optionnal

Column name representing the directions.

args: dict

Dictionnary with the folowing keys:

fitting distribution: str

Name of the fit to use, can be: Weibull, Gumbel, GPD or GEV.

method: str

Name of the estimation method, can be: pkd: Pickands’ estimator. pwm: PWM-method mom: Moment method ml : Maximum Likelihood method

threshold type: str

Method to find the peaks: percentile: using the th percentile value: using a treshold value

threshold value: float

Either a absolute value or percentile value depending on the threshold type

directional: bool

Can be True or False, to calculate stats for each direction. Needs drr variable

direction binning: str

Can be centered or not-centered depending if the directionnal are centered over 0

direction interval: int

Dirctionnal interval for the bins in degrees

minimum number of peaks over threshold: int minimum time interval between peaks (h): int display peaks: bool

True or False to display peaks over threshold

display CDFs: bool

True or False to display CFDs image

folder out: str

Path to save the output

time blocking: str

if Time blocking=='Annual',

Statistics will be calculated for the whole timeserie

if Time blocking=='Seasonal (South hemisphere)',

Statistics will be calculated for each South hemisphere seasons

if Time blocking=='Seasonal (North hemisphere)',

Statistics will be calculated for each North hemisphere seasons

if Time blocking=='Monthly',

Statistics will be calculated for each 12 months

>>> df=tf['test1']['dataframe'].Extreme.distribution_shape(mag='U',drr='drr',args={'directional':'On',Time blocking':'Annual'})
>>> 

Distribution shape

Filename	Scale	Shape