Aug 27, 2017 · 1 min read
Scientific Python (python, numpy, scipy, pandas, scikitlearn, etc. and several hundred packages) should be installed using a proper distribution, such as Anaconda3 from Continuum. An upgrade risk-free of all packages is simple as “conda update<blank> — all”. Anaconda even installs a compatible pip and it is very easy to build whole dedicated virtual environments, with their own version of Python. Also, in the case of Linux, you should avoid at all costs tinkering with the system’s Python (with sudo!…).
