Coursera, Big Data 4, Machine Learning With Big Data (week 3/4/5)
week 3 Classification
KNN :基本思想是 input value 类似,就可能是同一类的
Decision Tree
Naive Bayes
Week 4 Evaluating model
Over-fitting
怎么在Decision Tree 训练时避免 overfitting: Pre-Pruning 和 Post-Pruning
pre-pruning 两个停止条件:1. 某个node上的record数目小于一定量,比如 <20个, 2. 纯度到达一定数值,比如80%, 就不再split了.
怎么取 validation set
holdout 方法如下表示,为了解决training set 和validation set 可能distribution 不同,还有一个引申出来的repeated-holdout
除了 accuracy, error rate, F1, Confusion Matrix
Week 5 Regression, Cluster, Association
Association:
先create 1-item set,根据support 的threshold, 去掉低于 min support 的部分,然后再create 2-item set
转载请注明出处 http://www.cnblogs.com/mashuai-191/