1 / 3
文档名称:

几种监督式学习算法的比较.xls

格式:xls   大小:12KB   页数:3页
下载后只包含 1 个 XLS 格式的文档,没有任何的图纸或源代码,查看文件列表

如果您已付费下载过本站文档,您可以点这里二次下载

分享

预览

几种监督式学习算法的比较.xls

上传人:慢慢老师 2022/1/17 文件大小:12 KB

下载得到文件列表

几种监督式学习算法的比较.xls

文档介绍

文档介绍:Comparing Supervised Learning Algorithms (几种监督式学****算法的比较)
Algorithm(算法) Problem Type(问题类型) Results interpretable byComparing Supervised Learning Algorithms (几种监督式学****算法的比较)
Algorithm(算法) Problem Type(问题类型) Results interpretable by you?(预测结果是Classification还是Regression) Easy to explain algorithm to others?(解释的难易度) Average predictive accuracy(预测的平均准确性) Training speed(训练速度)
KNN(临近算法) Either Yes Yes Lower Fast
Linear regression(线性回归) Regression Yes Yes Lower Fast
Logistic regression(逻辑回归) Classification Somewhat Somewhat Lower Fast
Naive Bayes(朴素贝叶斯分类算法) Classification Somewhat Somewhat Lower Fast (excluding feature extraction)
Decision trees(决策树) Either Somewhat Somewhat Lower Fast
Random Forests(随机森林) Either A little No Higher Slow
AdaBoost(AdaBoost算法) Either A little No Higher Slow
Neural networks(神经网络) Either No No Higher Slow

Prediction speed(预测速度) Amount of parameter tuning needed (excluding feature selection)(需要调试的参数数目,不包括属性) Performs well with small number of observations?(是否进行少量的观察就能很好地工作?) Handles lots of irrelevant features well (separates signal from noise)?(是否能很好地处理无关的属性,将信号从噪声中分离出来?) Automatically learns feature interactions?(是否能够自动处理特征交互?) Gives calibrated probabilities of class membership?(是否提供了分类概率校准功能?) Parametric?(是否为参数式的学****Features might need scali