computer algorithm that use data to make educated guesses or dessisions.
'traditional algorithm' needs to be hard coded for improvement whereas 'Machine Learning model-algorithm' improves on its own with the help of input data.
For example,
spam filters use machine learning. 20 years ago, spam filters did not have many examples to learn from and were not good at identifying what is and isn’t spam. As more spam has arrived and been labeled as junk by human users, the machine-learning algorithms have gained more experience and become better at their job.
In this scenario, you own a shop that sells harnesses for avalanche-rescue dogs, and you’ve recently expanded to also sell doggy boots. Customers all seem to pick the correct harness sizes, but are constantly ordering doggy boots that are the wrong size. You know most customers buy harnesses and boots in the same transaction, which gives you an idea: perhaps you could approximate which doggy boots are the correct size, depending on the harness chosen. Then, you could warn customers if the boots they have selected are likely to be the correct size
before they make the purchase.
We’ll create a machine-learning model that does exactly this. Along the way, we’ll use this scenario to introduce some basic machine-learning concepts and demonstrate how to use them in a practical setting.