|
|
【答案】应助回帖
 商家已经主动声明此回帖可能含有宣传内容
★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★
yongcuo3: 金币+35, 翻译EPI+1, ★★★★★最佳答案 2016-03-10 12:12:16
在统计显著性 检验方面，事实上 前期 我们 已考虑对 两种方法 在7个数据集 上所得到的 性能指标 进行 Z测试，但做Z测试 必须 计算方差，而这7种数据集的 差异较大，求得的 方差较大，所以 用Z测试 意义不大。而进行t测试 也有 类似的问题。所以 我们在 未来的工作中，我们将 选择 更有 比较意义的 数据集，从而 更好地 进行 统计学上的 性能比较.
In aspect of statistical significance test, in fact we have already considered to perform Z test on performance indexes obtained by two methods based on 7 data sets. However, the calculation of variance is an inevitable process in Z test, and there is a large difference among these 7 data sets, so that the obtained variance is large, therefore it is of less significance to perform Test. In addition, there are similar problems in t test. As a results, we will select data set with comparative significance in our future work, so as to perform statistical comparison of performance. |
|
回复此楼