Introduction to Machine Learning with Python¶

by Andreas C. Müller and Sarah Guido

Introduction¶

Video course that supplements this book Code examples

Machine learning is about extracting knowledge from data. It is a research field at the intersection of statistics, artificial intelligence and compute science and is also often known as predictive analytics or statistical learning.

In the early days of "intelligent" applications, many used hardcoded if and else decisions to process data. For example for an email spam filter we could create a blacklist of words to find out if an email is spam or not. This is an example of an expert-designed rule system. This works well when humans have a good understanding of the process to model. However, there are two major disadvantages:

• The logic to make a decision is specific to a single domain and task. Changing the task only slightly could require and entire rewrite.
• Designing rules for such a system requires a deep understanding of how decision should be made by a human expert.

An example of where this approach will fail is detecting faces in images. A human know when there is a face, but he can't really describe a rule set for the system.