Fig. 8.


Download original image

Flowchart of the outlier detection with the possibility of physical outliers. First, we construct an autoregressive model of the time series. Then we flag all values whose difference to the model is greater than a given threshold n as possible outlier candidates. To account for the possibility of physical outliers, we relax this condition if proxies flag the same data as outliers.

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.