Python is interesting | machine learning should be learned like this

Python is interesting | machine learning should be learned like this


I don’t know how long it will take you to learn machine learning? The knowledge of Python crawlers and data analysis that we have introduced in the previous article, if you just do an introductory course, the problem is not big for an average of one month; but most of them feel that machine learning is difficult to learn and requires a good foundation in mathematics. Now graduates After a long time, I feel dizzy when I see the mathematical formula. Machine learning may go from getting started to giving up, and it takes a long time to learn.

You might go to Baidu or Google to search for how to learn machine learning. Most of the recommendations are these:

  • Video: Video of Mr. Wu Enda, National Taiwan University Li Hongyi
  • Books: Watermelon Books and Statistical Learning Methods

These are the essence of the big guys, and the quality of the content is very high. But are these really suitable for all novices? Not necessarily, the large number of derivation formulas in these books are actually difficult for those who are not determined.

Personal learning path

Based on this, I will share my personal experience of learning Python machine learning to everyone. Everyone should know the twenty-eight rule. The initial research on the twenty-eight rule was in economics. Now it is also used in the education industry. It means that learning 20% ​​of the most important knowledge points in an industry can solve 80% of industry problems.

For me, a forest sweeper, Python machine learning is mainly used for writing papers. I would ask myself, do you want to derive the entire machine learning algorithm? Actually, no, I just need to understand the algorithm and apply it to my forestry data.

For those who want to engage in machine learning in the future, this method is actually also practical. If you come to gnaw on the watermelon book from the beginning, it is very likely that your enthusiasm will be discouraged and there will be no motivation to learn. Then we will get started briefly, then come back to learn these knowledge points systematically, and read the books of these big guys, and you will get twice the result with half the effort.

So how easy is it to get started with machine learning? I have roughly mentioned it before, that is, simply understand the algorithm + apply the algorithm. What should I do? This is my previous learning method and path, which is only for your reference.

  • "Machine Learning Practical Combat", this book is a book about pure Python (but python2) code to implement machine learning. The introduction to the algorithm is very simple and clear, and there are not many formulas. If your code ability is limited, you can first understand its algorithm principles and programming steps, and you can write pseudo-code yourself.
  • The blog, although "Machine Learning Practical Combat" is simple and clear when introducing the principle of the algorithm, it is sometimes too simple to be particularly clear. At this time, you can read some blogs on Baidu, and some blogs are still well written.
  • sklearn practice, after understanding the principle of the algorithm, you can use our Python third-party library sklearn to practice it. The recommended book is "Python Machine Learning Basic Tutorial". In fact, you can find any book on sklearn practice. If you are good in English, you can directly read the official sklearn document.
Reference: Python is interesting|Machine learning should be learned like this-Cloud + Community-Tencent Cloud