SWXGModel

class swxg.SWXGModel(raw_data: DataFrame)

__init__(raw_data: DataFrame) → None

The base class to create, debias, fit, synthesize, and validate the stochastic weather generation model.

Parameters

raw_data: pd.DataFrame

Input dataframe to use as for stochastic weather generation

Attributes

raw_data: pd.DataFrame

Mirrors in the input raw_data parameter

data: pd.DataFrame

Temporal reformatting of raw_data so that each year, month, day have their own column in the DataFrame

resolution: str

Determined resolution of the observed data. Can be “monthly” or “daily”

precip_fit_dict: dict

Dictionary containing statistical information related to fitting of precipitation data

copulaetemp_fit_dict: dict

Dictionary containing statistical information related to fitting of copulae and temperature data

is_fit: bool:

Flag to confirm that this instance of the generator has been fit

fit(verbose: bool = True, validate: bool = True, dirpath: str = '', kwargs: dict = {}) → None

Local method that calls the fit_data function, which aptly fits the input weather data by returning extracted statistical parameters and reformatted data

Parameters

verbose: bool, optional

Flag for displaying precipitation and temperature fit statistics. Default: True

validate: bool, optional

Flag for producing figures to validate each step of the generator. Default: True

dirpath: str, optional

Path for where to save the validation figures. Default: “”

kwargs: dict, optional

Keyword arguments related to the fit. Leaving this empty sets the keyword arguments to their default values. Keywords are:

• gmhmm_min_states: int, default = 1

• gmhmm_max_states: int, default = 4

• gmhmm_states: int, default = 0

• max_convergence_attempts: int, default = 50

• ar_lag: int, default = 1

• copula_families: list[str], default = [“Independence”, “Frank”, “Gaussian”]

• figure_extension: str, default=”svg”

• validation_figures: list[str], default = [“precip”, “copula”]

format_time_resolution(data: DataFrame) → None

Function that separates the raw data’s datetime stamps to individual dataframe columns based on the input resolution

Parameters

data: pd.DataFrame

Input raw data to be used for the fitting

synthesize(n: int = 0, resolution: str = '', validate: bool = True, dirpath: str = '', kwargs: dict = {}) → DataFrame

Local method that calls the synthesize_data function, which aptly synthesizes weather through either observed fit or given statistical parameters

Parameters

n: int, optional

Number of years to synthesize. Default takes the same size as the number of years in the observed dataset

resolution: str, optional

The resolution to synthesize the data at. Leaving this empty sets the same resolution as the raw data

validate: bool, optional

Flag for producing figures to validate each step of the generator. Default: True

dirpath: str, optional

Path for where to save the validation figures. Default: “”

kwargs: dict, optional

Keyword arguments related to the fit. Leaving this empty sets the keyword arguments to their default values. Keywords are:

• validation_samplesize_mult: int, default = 10

• figure_extension: str, default=”svg”

Returns

synthesized_data: pd.DataFrame

The synthesized weather data at the desired time resolution