What is machinelearning_python?

This repository contains Python implementations of common machine learning algorithms, written primarily in Chinese. It is structured as a study resource that walks through the mathematics behind each algorithm and then shows the corresponding Python code. For each topic, there is also a version using scikit-learn, a popular Python library that provides ready-built versions of the same algorithms. The algorithms covered include linear regression (predicting a numerical output from input data), logistic regression (classifying inputs into categories), a neural network built using backpropagation (a method for training networks layer by layer), support vector machines (a technique for drawing decision boundaries between categories), K-Means clustering (grouping data points into clusters without pre-assigned labels), PCA or principal component analysis (reducing the number of dimensions in a dataset to simplify it), and anomaly detection (identifying data points that deviate significantly from the expected pattern). For each algorithm, the README presents the cost function (a measure of how wrong the model's predictions are), the gradient descent update rule (the step-by-step procedure for improving the model), and a Python code snippet that implements it. Screenshots of results are also included so a reader can see what the output looks like when each algorithm runs. The audience is students and developers learning how these algorithms work under the surface, rather than those who just want to call a library function. Most explanations are in Chinese, so readers who do not read Chinese will find the code samples and mathematical notation more accessible than the prose. The full README is longer than what was shown.