1. Overview of machine learning
Machine learning is rapidly changing our world. At the heart of artificial intelligence, we read almost every day how machine learning can change our everyday lives. Some believe it will lead us to a bizarre high-tech utopia; others believe we are heading towards a high-tech apocalypse, where we will wage a constant war against the robots and death squads that will steal our jobs.
But while pundits may love to discuss these hyperbolic futures, the more mundane reality is that machine learning is fast becoming a fixture of our everyday lives. Machine learning is quietly improving our lives as we make small but incremental improvements to how we interact with our computers and the world around us.
The process of machine learning is basically that users generate data which is collected, aggregated and fed into the model which ultimately creates a personalized experience for each user to improve the service.
Python is ideal for developing these applications. Python has a deep and active community of developers, many of whom also come from the community of scientists. This provides Python with a rich set of scientific computing libraries.
2. Machine Learning Workflow
Building machine learning applications is similar in many ways to standard engineering paradigms, but differs in one very important way: the need to process data as raw material.
The success of a data project depends heavily on the quality of the data you get and how it is processed. Since the use of data belongs to the field of data science, it is also helpful for us to understand the workflow of data science: the whole process should follow the sequence in the figure below to complete
six steps: acquisition, inspection and exploration, cleaning and preparation, modeling, Evaluation and final deployment.
During this process, it is also often necessary to go back to previous steps, such as checking and preparing data