1, learning and mastering some mathematical knowledge, advanced mathematics is the foundation of the foundation, all science and engineering need this foundation, data mining, artificial intelligence, pattern recognition and other data processing especially need the foundation of multivariate calculus operation.

2. Master the classic machine learning theory and algorithm. If you have time, you can build yourself a knowledge map of machine learning and try to master every classic machine learning theory and algorithm.

3. Master a programming tool, such as Python. On the one hand, Python is a scripting language, which is simple, can be written in Notepad, and can be run on the console after writing. In addition, Python is very efficient, which is higher than java, R and matlab. Although matlab has many packages, its efficiency is the lowest among the four.