|
Figure 5.7: Standard deviation versus mean relationship for the CCl4sn data. For each feature, the plot shows the empirical standard deviation sigma_i of the normalized and glog-transformed data on the y-axis versus the rank of the mean m_i on the x-axis. The red dots, connected by lines, show the running median of the standard deviatiqon. The rank of the smallest value is 1, the rank of the largest value is 43,628 (the number of features in CCl4sn). The rank scale is used for the x-axis in order to distribute the data evenly along the x-dimension. This allows for a better visual assessment of mean-variance trends compared to when the original scale is used.
|
|