Besides getting started, reptiles are also widely used in some companies, platforms and institutions that need data. It is very common to realize some commercial value by grabbing public data on the Internet. Of course, these players' crawlers are much stronger, and they need to deal with many problems, including routing, storage, distributed computing and so on, which are many times more complicated than Xiaobai's Huang Tu applet.
In addition to reptiles, Python is also widely used in Web programs. For example, the Zhihu you are using now has a tornado framework based on Python in the background of the main station and a tornado framework based on Python in the background of Douban. In addition to tornado (Tornado Web Server), Python commonly used web frameworks include FLASK (Welcome | FLASK (a Python micro-framework)), Django (a web framework for perfectionists with deadlines) and so on. Through the above framework, you can easily implement a Web program. For example, some friends I know have written their own blog programs through Python, including the previous ones. Zhihu.photo, just realized the background through Flask (I have stopped this website for copyright reasons). In addition to the above framework, you can also try to implement a Web framework yourself.
Artificial intelligence (AI) and machine learning artificial intelligence are a very hot direction now, and the AI craze makes the future of Python language full of infinite potential. Several influential AI frameworks released now are mostly implemented in Python. Why? Because Python is dynamic enough and has enough performance, this is the technical feature required by AI technology. For example, some websites based on Python's deep learning library, deep learning direction, machine learning direction and natural language processing direction are basically realized through Python.
Machine learning, especially the popular deep learning, most of its tool frameworks provide Python interfaces. Python has always had a good reputation in the field of scientific computing, and its concise grammar and rich computing tools are deeply loved by developers in this field.
Long before deep learning and Tensorflow became popular, there was scikit-learn in Python, which can easily complete almost all machine learning models. You can build a model by downloading a few simple lines of code from classic data sets. With tools such as Panda and matplotlib, it can be easily adjusted.
Deep learning frameworks such as Tensorflow, PyTorch, MXNet and Keras greatly expand the possibilities of machine learning. Using Keras to write a deep learning network for handwritten numeral recognition requires only dozens of lines of code, and with the help of the underlying implementation, a large number of resources including GPU can be conveniently called to complete the work.
It is worth mentioning that no matter what framework, Python is only used as a front-end description language, and the actual calculation is realized through the underlying C/C++. Python can easily introduce and use C/C++ projects and libraries, thus expanding its functions and performance. One of the important reasons why Python is widely used in the field of machine learning in such a large-scale calculation is to make developers pay more attention to logic and get rid of complicated work such as memory allocation.
The development efficiency of scientific computing Python is very high, and modules with high performance requirements can be rewritten in C and called by Python. At the same time, Python can abstract problems at a higher level, so it is also very popular in the field of scientific computing. Including scipy, numpy and other scientific computing third-party libraries, more convenient and friends with a certain mathematical foundation, but the general computer foundation.