|
监督学习,supervised learning
无监督学习,unsupervised learning
分类,classificat
回归,regression
降维,dimensionality reduction
聚类,clustering
特征向量,feature vector
编译语言,complied languages
解释型语言,interpreted languages
解释器,interpreter
布尔值,boolean
元组,tuple
算术运算,arithmetic operators
比较运算,comparison operators
赋值运算,assignment operators
逻辑运算,logical operators
成员运算,menbership operators
二分类,binary>
多分类,multiclass>
多标签分类,multi-lable>
线性分类器,linear> 系数,coefficient
截距,intercept
参数,parameters
随机梯度上升,stochastic gradient ascend(SGA)
预测结果,predicted condition
正确标记,true condition
混淆矩阵,confusion matrix
准确性,accuracy
召回率,recall
精确率,precision
随机梯度下降模型,SGD>
支持向量机分类器,support vector> 朴素贝叶斯,naive bayes
K近邻分类器,KNeighborsClassifier
无参数模型,nonparametric model
信息熵,information gain
基尼不纯性,gini impurity
集成,ensemble
单一决策树,decision tree
随机森林分类器,random forest> 梯度提升决策树,gradient tree boosting
平均绝对误差,mean absolute error(MAE)
均方误差,mean squared error(MSE)
极端随机森林,extremely randomized trees
随机回归森林,randomforestregressor
极端回归森林,extratreesregressor
核函数,kernal
scikit-learn
针对房价预测的回归预测能力排名,R-squared(用来衡量模型回归结果的波动可被真实值验证的百分比,也暗示了模型在数值回归方面的能力)
1,gradient boosting regressor,0.8426
2,extra trees regressor,0.8195
3,random forest regressor,0.8024
4,SVM regressor(RBF kernel),0.7564
5,KNN regressor(distance-weighted),0.7198
6,decision tree regressor,0.6941
7,KNN regressor(uniform-weighted),0.6903
8,linear regressor,0.6763
9,SGDregressor,0.6599
10,SVM regressor(linear kernel),0.6517
11,SVM regressor(poly kernel),0.4045
泛化力,generalization
正则化,regularization
过拟合,overfitting
留一验证,leave-one-out cross validation
交叉验证,K-flod cross-validation |
|