CR flagging is currently based on computing a weighted average of the flux over the intervals and then checking for outliers. This is prone to problems when few reads are available, since such estimate is biased by the outliers (i.e.the CRs) themselves, if they have not been identified yet.
One could explore the effects of using a median or a sigma-clipped average instead
CR flagging is currently based on computing a weighted average of the flux over the intervals and then checking for outliers. This is prone to problems when few reads are available, since such estimate is biased by the outliers (i.e.the CRs) themselves, if they have not been identified yet.
One could explore the effects of using a median or a sigma-clipped average instead