SymPy is a scientific computing library in python, which has a powerful scientific computing system, covering many fields from basic symbolic operations to computational mathematics, algebra, discrete mathematics, quantum physics and so on. Can complete polynomial evaluation, limit, equation, differential equation, series expansion, matrix operation and other calculation problems.

Although Matlab has a similar scientific computing ability, Python, with its simple grammar, ease of use and unusually rich three-party library ecology, personally thinks that it can solve various computing problems encountered in daily life more elegantly.

Numpy is an important software package for data analysis, machine learning and scientific computing. The operation and processing of vector moments are greatly simplified. Many data processing software packages of Python rely on Numpy as the core part of its infrastructure (such as Scikit-learn, Scipy, Pandas and tensflow). ).

Scipy is a scientific computing toolkit, which can handle interpolation, integration, optimization, image processing, data solution of ordinary differential equations, information processing and so on. It is based on Numpy. It can be used to calculate Numpy matrix effectively, make Numpy and Scipy work together and solve problems efficiently.

Sklearn is a machine learning package based on Numpy, Scipy and matplotlib. Its main functions are divided into six parts: classification, regression, clustering, data dimensionality reduction, model selection and data preprocessing, and its performance is also very good.

However, sklearn does not support languages other than deep learning and reinforcement learning, graph model and sequence prediction, python, PyPy and GPU acceleration.

Commonly used sub-modules are:

Statsmodels are used for fitting statistical models, parameter estimation, hypothesis testing, uncertainty evaluation and data exploration and visualization. Compared with sklearn, statsmodels focuses more on statistical reasoning, p-value and uncertainty evaluation. Commonly used sub-modules include:

Matplotlib is a drawing tool similar to matlab in python. In fact, matplotlib has a set of drawing interfaces completely in the form of MATLAB functions. In matplot.pyplot module, this set of functional interfaces is convenient for MATLAB users to transition to Matplotlib.

Seaborn is encapsulated on the basis of matplotlib, but seaborn is aimed at statistical drawing. Generally speaking, seaborn can meet 90% of the drawing requirements of data analysis.

Seaborn's goal is to make visualization a core part of exploring and understanding data. Its data set-oriented drawing function operates on data frames and data groups containing the whole data set, and performs necessary semantic mapping and statistical aggregation internally to generate an information map.

Seaborn can make heat maps, scatter charts, histograms, box charts, tree charts, hot spots and so on.

Pandas is built on Numpy array, which is specially used for processing tables and mixed data, while Numpy is more suitable for processing unified data array data.

References: