Data Science encompasses very many different fields.
Data in all its forms, both structured and unstructured; Big Data as well as relational data; archived data and streaming data, etc.
The 'data' source may be from business, science, social media platforms, and innumerable other sources. Data science is, therefore, multidisciplinary requiring methods, processes, algorithms to extract knowledge and insight which may fall into more complicated constructs such as AI, machine learning, and deep learning.
In order to tackle this very complicated situation, you need many different Python libraries. The ones shown here are no way complete as your problem may involved programs beyond the ones shown here.
Scikit -Essential for Machine Learning(ML)
----------------
PyTorch with at least Python 3.6 and Conda 4.6 (It is free) - ML and Natural Language Processing
----------------------------
Caffe --Deep Learning Framework
------------
TensorFlow - One-stop, open-source platform for ML
--------------------------
Theano - Evaluate mathematical expressions involving multi-dimensional arrays
------
Pandas -Data Analysis Library
-----------------------
Keras -Prototyping (Neural Network) deep learning
----------------
NumPy - Scientific Computing
[https://numpy.org/doc/stable/reference/index.html (for version v1.18 Manual)
https://github.com/SciSharp/Numpy.NET C# Bindings
--------------------
Matplotlib -Data Visualization
---------------------
SciPy -Technical and scientific computing
---------------
