First install miniconda if you haven't already.
Keras (Python package) is installed by installing Tensorflow. As of 2024, Keras is in transition from version 2 to version 3. To keep everything stable (and working with R), we're going to install version 2 (via Tensorflow version 2.13). It is necessary to use pip install within the conda environment, rather than conda install, because the conda repositories don't have later versions of the tensorflow packages for Windows.
For most people, you should be able to install everything following the official installation directions. This amounts to starting R and running this:
install.packages("keras")
keras::install_keras(method = "conda", python_version = "3.10")If you find this doesn't work for whatever reason, start up miniconda, and delete the conda environment (r-tensorflow) that was just created during the automated setup:
conda env remove --name r-tensorflowThen install python, tensorflow, and related packages manually. You need to pip install all packages at once, otherwise later package installs could update keras and tensorflow as dependencies, which will break things.
conda create --name r-tensorflow
conda activate r-tensorflow
conda install python=3.10 -c conda-forge
python -m pip install "tensorflow==2.13.*" tensorflow-hub tensorflow-datasets scipy requests Pillow h5py pandas pydotThe installation process is substantially the same but you're going to use conda and install a couple of extra packages (matplotlib, plotnine). You need to pip install all packages at once, otherwise later package installs could update keras and tensorflow as dependencies, which will break things.
conda create --name py-tensorflow
conda activate py-tensorflow
conda install python=3.10 -c conda-forge
python -m pip install "tensorflow==2.13.*" tensorflow-hub tensorflow-datasets scipy requests Pillow h5py pandas pydot matplotlib plotnineThen you need to set your IDE to start Python out of this environment.