用geom_smooth绘制置信区间显示不正确的y值

Question

我有关于rmse的各种值的数据，如下所示

> results
      X  r         rmse          mae       time
1     0  5 1.032451e+02 1.773904e+01 0.02531052
2     1  5 9.390353e+01 2.722062e+01 0.08579183
3     2  5 9.927726e+01 2.420522e+01 0.07660317
4     3  5 2.646419e+03 4.235173e+02 0.05406404
5     4  5 2.538567e+02 5.226082e+01 0.05190802
6     5  5 6.254017e+01 2.081937e+01 0.05366969
7     6  5 2.234349e+02 7.986439e+01 0.05228305
8     7  5 9.737024e+01 1.911705e+01 0.04921126
9     8  5 6.457242e+01 1.804460e+01 0.04793692
10    9  5 6.505051e+02 1.979597e+02 0.04743266
11   10  5 7.930607e-01 1.600373e-01 0.04627228
12   11  5 5.818517e+02 1.328608e+02 0.04759240
13   12  5 2.157449e+02 4.967495e+01 0.04678392

计算均方根和sd的均方根值（按r值分组）显示以下值

> ag <- aggregate(rmse ~ r, results, function(x) c(mean = mean(x), sd = sd(x)))
       r    rmse.mean      rmse.sd
1      5 7.859011e+02 3.562055e+03
2     10 5.565706e+03 2.668906e+04
3     15 4.930884e+03 2.480148e+04
4     20 1.417323e+04 5.822914e+04
5     25 1.630880e+07 1.627742e+08
6     30 1.818189e+04 5.416481e+04
7     35 2.361205e+04 1.035922e+05
8     40 9.020667e+04 5.576658e+05
9     45 1.808118e+05 1.547939e+06
10    50 2.550772e+04 1.660186e+05

但是，当我使用geom_smooth / stat_smooth绘制带状图时，我得到非常大的Y值。

> ggplot(results, aes(r, rmse)) + stat_smooth()