sklearn
w1a1ng
这个作者很懒,什么都没留下…
展开
-
sklearn-4.3.prepreocessing
4.3.1If a feature has a variance that is orders of magnitude larger than others, it might dominate the objective function and make the estimator unable to learn from other features原创 2017-03-23 23:04:50 · 241 阅读 · 0 评论 -
sklearn-2.5.Decomposing signals in components
2.5.12.5.1.1Principal component analysis(PCA)用n维垂直向量表达原特征们尽可能多的信息。whiten=True设置向量们各自的标准差都是1。高斯核的SVM和K-Means可能会用到。PCA可以对各个特征的贡献进行评分,这可以用在交叉验证当中。2.5.1.2Incremental PCAPCA的一个限制是必须将数据集全部加载进内存中,对于大数据集用Incre原创 2017-03-26 09:38:45 · 284 阅读 · 0 评论