Since Jupyter's datascience notebook uses Linux kernel,
R installation is slow and sometimes installation fails.
The most simple solution for it is to use conda to install the package.
Jupyter's datascience notebook have already mamba (C written version of conda),
so use mamba to install the package.
You can find R packages from the anaconda package repository.
For example, if you want to install finalfit package,
write RUN mamba install -c conda-forge r-finalfit -y.
-y option is required to install package without confirmation.
Alternative way is use install2.r which requires the use of rocker images.
See the following post for the detail.
You can check pre-installed packages in jupyter's datascience image in jupyter/r-notebook section.