原文由 yuan314159(yuan314159) 发表:。
下面是在Unscrambler的帮助文件中找到的说明,根据其描述验证集应该是Test matrix。
Leverage
Correction A method used as a first pass model check. This should not to be used as a final model validation method, as it an overly optimistic approximation.
Cross Validation
This method is used when either there are not enough samples available to make a separate test set, or for simulating the effects of different validation test cases, e.g. systematically leaving samples out vs. randomly leaving samples out, etc. See Cross validation setup dialog usage
Test matrix
This is also known as Test Set Validation, and uses independent samples that have not taken part in the calibration for validation. This allows one to define either a new matrix, of the same number of variables, or a defined range within a single matrix to be used as an independent check of model performance. Both X- and Y-matrices need to be defined in this case. This is the preferred method for validation and should be aimed for.
原文由 tearguodong(tearguodong) 发表:原文由 tearguodong(tearguodong) 发表:
刚开始使用unscrambler,有些看不懂interaction and square effect是什么意思呢?是交互作用?主成分分析要建立交叉验证吗?根据什么判别是不是离群值呢?大家有unscranbler9.7定性分析的资料吗?可以发给我吗?问题有点多,希望知道的大侠帮帮我。O(∩_∩)O谢谢
大侠帮帮忙,请问怎么把聚类分析的数据变成树形图呢?谢谢
原文由 ydlcsu(ydlcsu) 发表:原文由 tearguodong(tearguodong) 发表:原文由 tearguodong(tearguodong) 发表:
刚开始使用unscrambler,有些看不懂interaction and square effect是什么意思呢?是交互作用?主成分分析要建立交叉验证吗?根据什么判别是不是离群值呢?大家有unscranbler9.7定性分析的资料吗?可以发给我吗?问题有点多,希望知道的大侠帮帮我。O(∩_∩)O谢谢
大侠帮帮忙,请问怎么把聚类分析的数据变成树形图呢?谢谢
unscrmabler9.7应该是不可以弄成树形图,高版本的不清楚,没用过,或者换其他的软件吧。