NEWS.md

In spesenti/SWIM: Scenario Weights for Importance Measurement

SWIM 1.0.0 - current version on GitHub

Major changes: Additional functions and features

• Wasserstein distance

  • stress_wass():
    • A wrapper for the stress functions using the 2-Wasserstein distance
  • stress_RM_w():
    • a stressed model component (random variable) fulfills a constraint on its risk measure defined by a gamma function.
  • stress_RM_mean_sd_w():
    • a stressed model component (random variable) fulfills a constraint on its mean, standard deviation, and risk measure defined by a gamma function.
  • stress_HARA_RM_w():
    • a stressed model component (random variable) fulfills a constraint on its HARA utility defined by a, b and eta parameter and risk measure defined by a gamma function.
  • stress_mean_sd_w():
    • a stressed model component (random variable) fulfills a constraint on its mean and standard deviation.
  • stress_mean_w():
    • a stressed model component (random variable) fulfills a constraint on its mean.

• Functions

  • mean_stressed():
    • sample mean of chosen stressed model components, subject to the calculated scenario weights.
  • sd_stressed():
    • sample standard deviation of chosen stressed model components, subject to the calculated scenario weights.
  • var_stressed():
    • sample variance of chosen stressed model components, subject to the calculated scenario weights.
  • cor_stressed():
    • sample correlation coefficient of chosen stressed model components, subject to the calculated scenario weights.
  • cdf_stressed():
    • the empirical distribution function of a stressed model component (random variable) under the scenario weights.
  • rename_SWIM():
    • Get a new SWIM object with desired names.

• Features

  • stress():
    • A parameter "names" to all stress functions, which allows to name a stress differently than just "stress 1", "stress 2", etc.
    • A parameter "log" that allows users to inspect weights' statistics, including minimum, maximum, standard deviation, Gini coefficient, and entropy.
  • sensitivity():
    • A parameter "p" can be specified for the degree of Wasserstein distance.

Minor changes

• fix minor bug in summary().
• add base argument for quantile_stressed() and an error message if the input has wCol has dimension larger than 1.

SWIM 0.2.2 - current version on CRAN

Major changes: Additional functions and features

• plot_quantile():

  • the function plots the empirical quantile of model components, subject to scenario weights.

• plot_weights():

  • the function plots the scenario weights of a stressed model against model components.

• stress_moment():

  • add parameter "normalise" that allows to linearly normalise the values called by nleqslv.
  • the function prints a table with the required and achieved moments and the absolute and relative error.

• stress_VaR_ES():

  • add parameter "normalise" that allows to linearly normalise the values before uniroot is applied.

Minor changes

• fix bug in merging different stress objects.

SWIM 0.2.1

Minor changes

• add vignette
• fix bug in merge().
• fix bug in sensitivity().

SWIM 0.2.0

Major changes

Additional functions and data sets

• VaR_stressed():

  • the function calculates the VaR of model components, subject to scenario weights.

• ES_stressed():

  • the function calculates the ES of model components, subject to scenario weights.

• credit_data:

  • a data set containing aggregate losses from a credit portfolio, generated through a binomial credit model.

Amendments to functions

• stress_VaR():

  • amendment to the calculation of scenario weights when the specified VaR cannot be achieved.
  • returns a message if the achieved VaR is not equal to the stressed VaR specified.
  • specs of the SWIM object contains the achieved VaR
  • allowing for stressing VaR downwards

• stress_VaR_ES():

  • amendment analogous to the stress_VaR().
  • returns a message if the achieved VaR is not equal to the stressed VaR specified.
  • specs of the SWIM object contains the achieved VaR
  • allowing for stressing VaR and ES downwards

Minor changes

• stress():

  • parameter x can have missing column names.

• stress_moment():

  • additional parameter show; if TRUE (default is FALSE), the result of nleqslv() is printed.

spesenti/SWIM documentation built on Jan. 15, 2022, 11:19 a.m.