Question

If I use from sklearn import * or from skleanr import datasets, then I can use datasets in the following way: iris = datasets.load_iris().

However, import sklearn and import sklearn as sk do not work as I expect. For example I cannot use sklearn.datasets.import_iris() or sk.datasets.import_iris(). Do I misinterpret the import syntax? And it it is the case, what is the correct way to use import.

Answer 1

No, you are not misinterpreting it. It’s the package structure of this particular project.

When you import sklearn, you import a special python file __init__.py in a directory sklearn, that has inside of it another package called datasets. But if sklearn itself doesn’t import the nested package into it’s __init__.py file, you cannot use attribute traversal to get to that nested package.

The solution is to import the nested package explicitly, yourself:

import sklearn.datasets