Changes in version 2.1.0 (2024-09-09)                  

The 'build_aggtree' function now requires a different set of arguments
that allow better control on training-honest sample split.

                        Changes in version 2.0.3                        

Updated vignettes. Improved LATEX output.

                 Changes in version 2.0.2 (2023-09-20)                  

Fixed a bug that occurred when using only one covariate.

                 Changes in version 2.0.1 (2023-08-08)                  

Now the inference_aggtree function can estimate asymmetric
bias-corrected and accelerated 95% confidence intervals for the GATEs
based on their bootstrap distributions.

Also, improved automatic check of leaf composition. Functions that
estimate the GATEs now:

  - Raise an error if one or more leaves contain only one observation;
  - Raise an error if differences in mean outcomes is the chosen
    estimator and one or more leaves contain only treated or only
    control units.

                 Changes in version 2.0.0 (2023-02-21)                  

Updated hypothesis testing procedure. Improved LATEX output.

                 Changes in version 1.0.0 (2023-02-09)                  

First public release