减少SPSS中的特征数量
我有一个包含 200 多个特征的数据集,我想减少数量,以免高估结果的预测。
有谁知道SPSS中是否有任何选项可以计算目标值(Y)和自变量(X)之间的互信息或任何其他方法来检查哪些变量相关,哪些变量不相关?
谢谢你!
I have a dataset with more than 200 features and I would like to reduce the number in order not to overestimate the prediction of the outcome.
Does anyone know whether there is any option in SPSS to calculate mutual information between the target value (Y) and the independent variables (X) or any other method to check which variables are relevant and which are irrelevant?
Thank you!
如果你对这篇内容有疑问,欢迎到本站社区发帖提问 参与讨论,获取更多帮助,或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱,如果其他用户或者作者回复了您的评论,将不能在第一时间通知您!
发布评论
评论(1)
我没有看到在这种情况下使用“功能”一词,但我认为您需要 Principal成分分析。
然而,在不知道自己在做什么的情况下进行统计是得到毫无意义的数字的好方法;我建议你咨询统计学家。
I've not seen the term "features" used in this context, but I think you are in need of Principal Component Analysis.
However, doing statistics without knowing what you are doing is a good way to make meaningless numbers; I suggest you consult a statistician.