警告:r中的k-means聚类在10次迭代中没有收敛

Question

我试图通过使用Gap统计来找到最优的簇数。我已经增加了iter。最大到50:

差距统计h5> 是，覆盖问题没有解决，我仍然得到以下警告:

Warning: did not converge in 10 iterations
......Warning: did not converge in 10 iterations
.Warning: did not converge in 10 iterations
....Warning: did not converge in 10 iterations
.Warning: did not converge in 10 iterations

Answer 1

看起来您遇到的问题是您包含了iter。Max =50在实验室而不是fviz_nbclust函数。试试这个,

set.seed(123)
fviz_nbclust(data.s[, -c(1)], kmeans, nstart = 25,  method = "gap_stat", nboot = 1000, iter.max=50)+
  labs(subtitle = "Gap statistic method")