Changelog • PTE

PTE 2.0.1

Added the package citation for Kapelner et al. (2021), Evaluating the Effectiveness of Personalized Medicine With Software.

Modernized the package using Rcpp and ggplot2.
Profiled the package and eliminated repetitive overhead.
Added a comprehensive test suite and forward-facing API argument checks.
Implemented logistic regression and Weibull survival estimation in C++ using RcppEigen for substantial speedups.
Improved parallelization using mirai on Windows and forking on Unix-like systems.
Added opt-in, machine-specific compilation flags for local performance builds while keeping distributed builds portable by default.
Changed the default to the m-out-of-n bootstrap for nonsmooth policy-value estimands and corrected the rate adjustment using sqrt(m / n). The default is m = n^(3/4), following Politis, Romano, and Wolf (1999).
Added select_optimal_m_prop(), implementing the minimum-volatility grid method recommended by Politis, Romano, and Wolf (1999) and Bickel and Sakov (2008).
Fixed a bug in the difference_function branch.
Fixed documentation rendering issues.

CRAN release: 2019-01-30

Calculated and printed basic and studentized bootstrap confidence intervals by default.
Fixed warnings.

CRAN release: 2017-10-10

Moved the plot parameter into a separate function for ease of use.
Removed the num_cores parameter. Use the standard options(mc.cores = ...) setting instead.
Corrected the paper examples on the GitHub page.

CRAN release: 2017-06-12

Added support for survival endpoints.
Added support for incidence endpoints: odds ratio, risk ratio, and probability difference.
Improved arguments and defaults for PTE_bootstrap_inference().
Added customization for non-standard metrics for all endpoints.
Improved performance and documentation.

CRAN release: 2014-05-01