Installation

Install from PyPI

The standard version can be intalled with:

pip install pyESD

The same version can be installed in editable mode with the e flag:

pip install -e pyESD

Install from Github

The updated version (in development) can be clone from github:

git clone git@github.com:Dan-Boat/PyESD.git

and use pip to install (in the clone folder) .

pip install

It is recommended to install the package in an isolated environment. Virtualenv or conda can be used to create a new environment. The package requires some dependencies that can be installed through the distro. If failed to be install through pip installation, the following modules would require manual installation.

Common issues

Installing and Using Cartopy for Data Visualization

Visualizing large-scale datasets with cartopy can be a bit challenging. However, if you plan to use cartopy, keep in mind that it’s not necessary for any of the downscaling routines. Here’s a step-by-step guide to set it up:

Create a Cartopy Environment: First, create a new environment specifically for cartopy:
```
conda create --name cartopy_env
```
Install Cartopy: Once the environment is set up, install cartopy using the conda-forge channel:
```
conda install -c conda-forge cartopy
```
Clone Cartopy Environment: When creating an environment (“env_name”) for installing pyESD, you can clone the previously created cartopy environment to include its settings:
```
conda create --name env_name --clone cartopy_env
```

Keep in mind that cartopy has specific dependencies that need to be installed on your system. For more detailed information about these dependencies and how to set up cartopy, you can refer to the cartopy website: Cartopy Documentation. This will provide you with a comprehensive understanding of the setup process.

Dependencies

sklearn (pip install -U scikit-learn)
xarray (conda install -c conda-forge xarray dask netCDF4 bottleneck)
pandas
seaborn
tensorflow (pip install tensorflow)
matplotlib
netCDF4
eofs (pip install eofs)
cartopy (conda install -c conda-forge cartopy) (not required for the modelling routines), optional
xgboost (pip install xgboost)
scikit-optimize (pip install scikit-optimize)