Contents

3.Machine Learning

Contents

3.Machine Learning#

3、机器学习(Machine Learning)#

3.1 综述概览(General)#

(1) 概念、输入和属性(Concepts,Inputs & Attributes)#

1.分类变量(Categorical Variables)#

2.有序变量(Ordinal Variable)#

3.数值变量(Numerical Variable)#

(2) 代价函数和梯度下降(Cost functions and Gradient Descent)#

(3) 过拟合/欠拟合(Overfitting/underfitting)#

(4) 训练、验证和测试数据(Training,Validation and test data)#

(5) 精度和召回(Precision vs Recall)#

(6) 偏差和方差(Bias & Variance)#

(7) 优化提升(Model Lift)#

3.2 方法(Method)#

(1) 监督学习(Supervised Learning)#

1.回归(Regression)#

a.线性回归(Linear Regression)#

b.泊松回归(Poisson Regression)#

2.分类(Classification)#

a.正确分类率(Classification Rate)#

b.决策树(Decision Trees)#

c.逻辑回归(Logistic Regression)#

d.朴素贝叶斯分类器(Naive Bayes Classifiers)#

e.K-最近邻(K-Nearest Neighbour)#

f.支持向量机(Support Vector Machine)#

(2) 无监督学习(Unsupervised Learning)#

1.关联规则学习(Association Rule Learning)#

a.先验算法(Apriori Algorithm)#

b.ECLAT算法(ECLAT Algorithm)#

c.频繁模式树(FP-Tree,Frequent-Pattern tree)#

2.降维(Dimensionality Reduction)#

a.主成分分析(PCA,Principal component analysis)#

b.随机投影(Random Projection)#

c.非负矩阵分解(NMF,Non-negative matrix factorization)#

d.t-分布式随机邻域嵌入(T-SNE,t-Distributed Stochastic Neighbor Embedding )#

e.一致流形近似和投影(UMAP,Uniform Manifold Approximation and Projection)#

(3) 集成学习(Ensemble Learning)#

1.提升方法(Boosting)#

2.装袋算法(Bagging,Bootstrap aggregating)#

3.堆叠(Stacking)#

(4) 强化学习(Reinforcement Learning)#

Q-Learning#

3.3 用例(Use Cases)#

(1) 情绪分析(Sentiment Analysis)#

(2) 协同过滤(Collaborative Filtering)#

(3)标注(tagging)#

(4)预测(Prediction)#

3.4 工具(Tools)#

(1)I重要库(mportant Libraries)#

1.Scikit-Learn#

2.Spacy(NLP)#