Machine learning--initial knowledge and practical choice

As the beginning of the basic learning of machine learning, record the knowledge and practice choice of machine Learning Foundation.

Bibliography:

Machine learning Combat

An electronic version of both Chinese and English PDF files and code:here (azkx) is also available. I'm not going to tell you. Ctrl + A will have other discoveries ~

1. Basic

For machine learning related methods, the most intuitive understanding is that it can be divided into supervised and unsupervised two categories, you can use the mind map to visually Express:

(Use Mindnode to draw mind map, source file here)

Actually in the actual use, the classification method and the regression method demarcation is not obvious, sometimes also may combine to use.

The specific development steps have roughly these processes:

①, collecting and acquiring data

②, preparing input data

③, analysis preprocessing input data

④, training algorithm--unsupervised algorithm without training process, core content focused on ⑤

⑤, test algorithm

⑥, using algorithms

2. Practice

Many of the current machine learning practices are implemented on Python, so it's natural to choose Python.

The Python third-party libraries that may be used are:

Numpy, Scipy, Matplotlib, Pandas, Statsmodels, Scikit-learn, Keras, Gensim, etc.

Machine learning--initial knowledge and practical choice