Most people need to learn python when they get started with machine learning and artificial intelligence, so many people buy a book on python from entry to mastery, or just buy a book on basic python syntax. I started the same way, but in the process of learning, I found that even after reading the basic syntax of python, I still would not apply python to machine learning.
After tossing and turning, I felt that my theoretical foundation of machine learning was not enough, so I bought Li Hang's statistical learning method and Zhou Zhihua's machine learning, and combined with the research direction of the laboratory, I learned and supplemented the theoretical knowledge of machine learning step by step, but later found out , Usually the task of the laboratory is to implement basic machine learning algorithms (I was familiar with matlab before, and I have been using matlab to implement machine learning algorithms), and did not teach us how to use each machine learning algorithm, and how to call it if it is written in a program. Which functions and which parameters to call.
Later, after participating in Ali's Tianchi competition and some data mining competitions, I slowly learned that these basic machine learning algorithms can be called in various packages provided by python. At this time, I felt that the python I learned really used machine learning. The above, but I don’t know how to systematically learn these calling methods and how to adjust parameters. Later, I also learned from the learning methods of some big guys. You can
find
the
jupyter notebook code of the machine learning package you want to learn on github. There are explanations and code examples, which are quick to learn and easy to use.
Summarizing the method and mental process of learning machine learning in the past one year from research to now:
(1) First review some basic mathematical theories, such as linear algebra, probability theory, etc.
(2) Learn the theoretical basis of machine learning and preferably implement it by yourself, which is conducive to the understanding of the algorithm. Recommended python algorithm implementation link, you can refer to learning:
Machine learning algorithm implemented by python
, recommend two related theoretical books, one is Zhou Zhihua's machine learning, and the other is Li Hang's statistical learning method.
(3) Learn some basic syntax of python. It is recommended to use python language when implementing machine learning algorithms. I was lazy to use matlab at that time, which is not good!
(4) At this time, you must pay attention, don’t think that you can do machine learning by learning the basic syntax of python (of course, God can), you also need to learn to use the machine learning package, which encapsulates various machine learning algorithms, so learn With the basic syntax of python, this is just the beginning.
Next, you can learn various machine learning packages through the magic I said - jupyter notebook. For example, sklearn, pandas, numpy, matplotlib are commonly used. Searching for machine learning packages on github has jupyter notebook written by the great god.