1. Data science five commonly used Python library
Numpy
- N-dimensional array (matrix), fast and efficient vector math
- Efficient index, does not need to cycle
- Free open source cross-platform, operating efficiency and sufficient C / Matlab comparable
Scipy
- It depends on Numpy
- Designed for science and engineering
- Achieve a variety of commonly used scientific computing, such as: linear algebra, Fourier transform, signal and image processing
Pandas
- Structured data analysis tool (Numpy dependent)
- Offers a variety of advanced data structures: Time-Series, DataFrame, Panel
- Powerful indexing and data processing capabilities
Matplotlib
- Python 2D graphics areas most widely used suite
- The basic drawing functions can replace Matlab (scatter plots, graphs, bar charts, etc.)
- You can draw beautiful 3D drawing by mplot3D
Scikit-Learn
- Machine Learning Python module
- Built on scipy, it provides a common machine algorithms: clustering, regression
- Easy to learn API interface
2. Based on the review of the mathematical matrix operations
basic concept
- Matrix: a matrix array, ie two-dimensional array. Wherein the vector and scalar matrix is a special case.
- Vector: is a matrix 1xn or nx1
- Scalar: 1x1 matrix
- Array: N-dimensional arrays, matrix extension
Special Matrices
- 1 0 Full Full matrix: values are 0 or 1
- Matrix: multiplying a diagonal equal to a diagonal, any matrix multiplication and matrix are equal to the original matrix, nxn.
Matrix addition and subtraction
- Adding, subtracting two matrices must have the same columns and rows.
- Columns and rows corresponding element addition and subtraction.
An array multiplication (dot)
- Multiplication array (dot) is a multiplication between the corresponding elements
