old sklearn version in Jupyter Notebook

问题:

I update sklearn version by terminal with

conda install scikit-learn=0.18

if I list with conda list scikit-learn

# packages in environment at /Users/Claudia/anaconda:
scikit-learn              0.18.1              np111py27_1
scikit-learn              0.18.1                    <pip>

but if I run in notebook

print('The scikit-learn version is {}.'.format(sklearn.__version__))

the result is

The scikit-learn version is 0.17.1.

How can I solve and update also the version in Jupyter Notebook?

回答1:

to update the version used by Jupyter you need to open terminal by Jupiter interface

and run command from here

 conda update scikit-learn

the mistake was the use of the system terminal.

回答2:

I solved this by python -m pip install scikit-learn --upgrade
check Python loading old version of sklearn