Best Python machine learning library?

I'm looking for a Python library I can use to teach a program the difference between a positive, negative and neutral statement.

I don't really want to use NLP, I purely want to be able to tell it this is positive, this is negative and after enough test cases it will be able to decide itself.

So just wondering what you have used and how you found it. Code examples or GitHub links would be helpful.

For best Machine Learning library in python, you can look into scikit-learn. As you want to do text processing, I have worked on NLTK and found it pretty good. Your project or problem at hand is sentiment analysis. So,there would be a combination of nltk and scikit-learn. For documentation of scikit-learn - http://scikit-learn.org/stable/index.html For NLTK - http://www.nltk.org/