bonsai_ipcc package
Subpackages
- bonsai_ipcc.agriculture package
- bonsai_ipcc.industry package
- bonsai_ipcc.ppf package
- bonsai_ipcc.waste package
Submodules
bonsai_ipcc.core module
- class bonsai_ipcc.core.IPCC(_external_metadata=None, _ext_functions=None)[source]
Bases:
object- create_sample(dfs, constraints=None)[source]
Add sample to parameter tables.
For parameter tables in which values are related and constraint, Dirichlet distribution is used for samples. For other parameter tables, sample is determined independently using the best fit distribution.
Argument
- dfsdict
key is name of paramter, value is df
- constraints: dict
- key is name of parameter, value is dict with
group_py: [“year”, “region”, “product”]
- sizeint
sample size
- returns:
revised parameter tables with sample array
- rtype:
dict of dfs
- get_metadata(volume, chapter, parameter)[source]
Get the metadata of a parameter.
Argument
- volumestring
volume name
- chapterstring
chapter name
- paramaterstring
parameter name
- returns:
VALUE – metadata pointing to the source in the IPCC pdf documents (year, volume, chapter, page, equation)
- rtype:
dict
bonsai_ipcc.log_setup module
bonsai_ipcc.sample module
- bonsai_ipcc.sample.create_sample(dfs, constraints=None, size=1000)[source]
Add sample to parameter tables.
For parameter tables in which values are related and constraint, Dirichlet distribution is used for samples. For other parameter tables, sample is determined independently using the best fit distribution.
Argument
- dfsdict
key is name of paramter, value is df
- constraints: dict
- key is name of parameter, value is dict with
group_py: [“year”, “region”, “product”]
- sizeint
sample size
- returns:
revised parameter tables with sample array
- rtype:
dict of dfs
bonsai_ipcc.uncertainties module
Chapter 3 (Uncertainties) of Volume 1 (General Guidance and Reporting)
Min and max values provided by the IPCC guidelines are considerred to be 2.5th and 97.5th percentile, respectively (p 3.9).
Two ways of uncertainty propagation are implemented: analytical error propagation and Monte Carlo simulation. If analytical error propagation is choosen, asymmetric parameter ranges are transferred by a simple rule of thumb into symmetric ranges. If Monte Carlo, propability distributions can be determined automatically in regard to the properties of the parameter (by “check”).
- bonsai_ipcc.uncertainties.analytical(min95, max95)[source]
Creates ufloat value to be used in error propagation. Simple asumption of normal distribution.
Argument
- minfloat
2.5th percentile
- maxfloat
97.5th percentile
- returns:
VALUE
- rtype:
ufloat
- bonsai_ipcc.uncertainties.monte_carlo(min95, max95, default=None, abs_min=None, abs_max=None, size=1000, distribution='check')[source]
Creates numpy array of random numbers to be used in monte carlo simulation. Choices of different distributions possible.
Argument
- minfloat
2.5th percentile
- maxfloat
97.5th percentile
- defaultfloat
default value (mean)
- abs_minfloat
theoretic lower bound of parameter
- abs_maxfloat
theoretic upper bound of parameter
- distributionstr
Specifies type of density distribution. ‘normal’, ‘lognormal’, ‘truncnormal’, ‘uniform’, ‘check’ for automatized choice
- sizeint
Number of random numbers to be generated.
- returns:
VALUE
- rtype:
numpy.array
- bonsai_ipcc.uncertainties.one_side_trunc(mobs, sobs)[source]
Helper function for adjusting mean and sd for one-sided truncated normal distribution.
Given observed mean m > 0.0 and std-dev, 0.0 < s < m, return parameter mean and std-dev of one-sided truncated Gaussian distribution X, with X > 0.0
Rodrigues 2015, Maximum-Entropy Prior Uncertainty and Correlation of Statistical Economic Data: Supplementary Informatiom.
Argument
- mobsfloat
observed mean
- sobsfloat
observed standard deviation
- returns:
VALUES – adjusted mean, adjusted standard deviation
- rtype:
float, float
- bonsai_ipcc.uncertainties.sample_dirichlet(default, max95, abs_max, size=1000)[source]
Calculate samples for interconnected parameters.
Argument
- defaultlist
list of default values
- max95list
list of 97.5 percentiles
- abs_maxlist
list of absolute maximum
- sizeint
sample size
- returns:
each parameters sample in a row
- rtype:
np.array
- bonsai_ipcc.uncertainties.two_side_trunc(mobs, sobs, vmin, vmax)[source]
Helper function for adjusting mean and sd for two-sided truncated normal distribution.
Given observed mean m, std-dev s, lower and upper bounds, vmin and vmax, (all floats) such that: vmin < mobs < vmax, and 0.0 < sobs < min(mobs - vmin, vmax - mobs)
Argument
- mobsfloat
observed mean
- sobsfloat
observed standard deviation
- vminfloat
absolut minimum
- vmaxfloat
absolut maximum
- returns:
VALUES – adjusted mean, adjusted standard deviation
- rtype:
float, float