Python-based machine learning (1) - environment configuration
01. Basic introduction
It cannot be said that the current machine learning is very powerful, but it can be said that the role of the current machine learning in real life is also getting bigger and bigger, and in the future, the society's demand for talents in this area will also grow older. For machine learning, everyone who has heard of it basically knows how to use Python to learn. It has great advantages. You can make full use of the concept of Python without reinventing the wheel and using already written packages. We are optimizing on this basis. That's it.
The package used for machine learning in Python is scikilearn-learn , which relies on the Numpy, SciPy and matplotlib libraries. Commonly used functions in Scikit-learn are:
| .. | Application | Algorithm |
|---|---|---|
| Classification | Anomaly detection, pattern recognition, etc. | KNN, SVM, etc. |
| Clustering | Graph segmentation, group segmentation, etc. | K-Means, spectral clustering, etc. |
| Regression | Price forecasts, trend forecasts, etc. | Linear regression, SVR, etc. |
| 降维 （Dimension Reduction） | visualization | PCA, NMF, etc. |
Recommended books
1. "Machine Learning" - Zhou Zhihua
2. "Pattern Recognition and Machine Learning" - Christopher Bishop
02. Environment configuration
The environment of this series of blogs is the
installation of Numpy, Scipy, matplotlib, sklearn (also the order of installation) of the win10 system. The file address to be installed: https://www.lfd.uci.edu/~gohlke/pythonlibs/ , download according to the version in your computer Open dos
in the downloaded path, and fill in the command: pip install numpy-1.14.2 +mkl-cp35-cp35m-win_amd64.whl Waiting for installation. Install the Scipy and matplotlib packages in the same way; download your own scikit-learn package for the environment at https://pypi.org/project/scikit-learn/#files and continue to install it in the previous way. Import the package on the python command line in dos to see if the installation is successful
