accuracy: Tools for testing and improving accuracy of statistical results
This is a suite of tools designed to test and improve the
accuracy of statistical computation, including: Summarization
of the sensitivity of linear and non-linear models (lm, glm,
mle, nls) to measurement and numerical error; Sensitivity
analysis of dozens of models as run through Zelig; A
generalized cholesky method for correcting non-invertable
Hessians; Tests for the global optimality of non-linear
regression and maximum likelihood results; Tools for obtaining
true random numbers using entropy collected from the system
and/or entropy servers on the internet; A method for converting
floating point numbers to normalized fractions; Benchmark data
for checking the accuracy of basic distribution functions.
Downloads:
Reverse dependencies: