课程: Advanced Predictive Modeling: Mastering Ensembles and Metamodeling

今天就学习课程吧！

今天就开通帐号，24,600 门业界名师课程任您挑！

Is Occam's Razor always true?

Is Occam's Razor always true? - SPSS教程

课程: Advanced Predictive Modeling: Mastering Ensembles and Metamodeling

开始一个月试用 购买团队方案

Is Occam's Razor always true?

“

- [Instructor] If you have any formal traditional training in statistics, you've probably been coached to always favor the more parsimonious model. We're cautioned against overly complex models, so should we be worried about ensembles? Pedro Domingos has written an influential paper on this very subject, so it's worth pausing for a moment to reflect on what he concluded. By the way, when folks talk about Occam's Razor, they're referring to a quote from the Middle Ages that basically says simpler is better. So Domingos says that there's two senses in machine learning in which we tend to prefer the simpler model. And he distinguishes between the two. He doesn't dispute that the first is that we should prefer simpler because simplicity is desirable in itself. He doesn't dispute this. And frankly, it's often the case. For instance, I would identify two reasons why we might prefer the simpler model. One, frequently our model must be transparent. Regulation is a common example of this. Car…

内容