常见机器学习算法的Python实现

项目的地址在 https://github.com/WiseDoge/ML-by-Python

解决依赖

• NumPy

使用方法

所有的算法都重载了fit(train_x, train_y), predict(test_x)接口，使用方法类似sklearn。
例如：

from mlearn.knn import KNeighborClassifier
import numpy as np
clf = KNeighborClassifier(n_neighbors=3)
train_x = np.array([[1, 1], [0.1, 0.1], [0.5, 0.7], [10, 10], [10, 11]])
train_y = np.array(['A', 'A', 'A', 'B', 'B'])
test_x = np.array([[11, 12], [12, 13], [11, 13], [0.05, 0.1]])
clf.fit(train_x, train_y)
print(clf.predict(test_x))
# 输出['B' 'B' 'B' 'A']

from mlearn.naive_bayes import NaiveBayesClassifier
train_x = [["1", "S"], ["1", "M"], ["1", "M"],
           ["1", "S"], ["1", "S"], ["2", "S"],
           ["2", "M"], ["2", "M"], ["2", "L"],
           ["2", "L"], ["3", "L"], ["3", "M"],
           ["3", "M"], ["3", "L"], ["3", "L"]]
train_y = ["-1", "-1", "1", "1", "-1",
           "-1", "-1", "1", "1", "1",
           "1", "1", "1", "1", "-1"]
clf = NaiveBayesClassifier()
clf.fit(train_x, train_y)
print(clf.predict([["2", "S"]]))
# 输出['-1']

内容

• K近邻算法
• 感知机
• 朴素贝叶斯分类器
• 决策树(Decision Tree)
• 随机森林(Random Forests)
• 支持向量机(Support Vector Machine)
• 线性回归(Linear Regression)
• 逻辑斯蒂回归(Logistic Regression)
• Bagging算法
• 神经网络(BP算法)
• 隐马尔科夫模型(HMM)
• K-Means聚类算法
• LVQ聚类算法
• 主成份分析法(PCA)

说明

本项目中的算法是作者在学习机器学习的过程中所做的练习，其中不免有些错误，如果有发现的错误，欢迎批评指正。