Skip to main content
. Author manuscript; available in PMC: 2016 Oct 1.
Published in final edited form as: Comput Speech Lang. 2015 Jan;29(1):132–144. doi: 10.1016/j.csl.2014.02.001

Figure 2.

Figure 2

Distribution of pathology labels (I/NI) and EWE scores in each cluster for the train set. A total of 52 clusters were created. The histogram indicates the percentage of I annotations in clusters. The red line plot indicates the EWE mean and standard deviation. (Clusters sorted by average EWE score.)