The installation of GDAL for Python used to be quite complicated, requiring you to fiddle with system settings and path variables. It is still possible to install GDAL in various ways, however, we recommend that you use either Anaconda3 or conda, as this is the quickest and easiest way to get started. Other options are using pip install, or using an online repository such as http://gdal.org or Tamas Szekeres Windows binaries (http://www.gisinternals.com/release.php).

However, this might be a little more involved than the options described here. The catch with installing GDAL is that a particular version of the library (that comes in the C language and is installed in a separate system directory from your local Python files) has an accompanying Python version, and needs to be compiled in order for you to use it in Python. In addition, GDAL for Python is dependent on some extra Python libraries that come with an installation. While it is possible to use multiple versions of GDAL on the same machine, the recommended approach here is to install it in a virtual environment, using Anaconda3, conda, or pip installations. This will keep your system settings clean of additional path variables or stop things from working.