Data Science and Python
Deep Learning/Artificial Intelligence
google: tensorflow
facebook: pytorch+caffe2
Amazon: mxnet+gluon
Baidu: paddlepaddle
Some packages that are very easy to use : Keras Tflearn tensorlayer
machine learning
numpy scipy pandas scikit-learn xgboost/LightGBM
Big Data
Big data as a whole is dominated by languages such as Java/scala, but there are many ways of python that can be used seamlessly. From the early hadoop ecology, you can use python scripts to complete Map-Reduce tasks through hadoop streaming, and now everyone is in The widely used spark also has a python interface, and it is not much different from other languages in terms of efficiency.
reason
Time is very precious. "Algorithm engineers" and "data scientists" need to use data-driven methods to solve various problems in real life, but it is not a very efficient way to spend a lot of time on development. Ideally, you put The main focus is on analyzing business scenarios and data, finding better and more appropriate ways to solve the problem itself, and quickly coding, verifying or developing services.