Outlier Detection and Treatment

[#1]

I published a paper in the SESUG 2020 conference on the topic of outlier detection and treatment, and I am sharing it with the SAS community.

I describe the effects that outliers have on parameter estimates, univariate and multivariate detection methods, and local outlier factor (LOF) and local outlier probability (LoOP) detection methods. I discuss outlier treatment methods such as excision, using binary indicator variables, Winsorization, and transformations. I wrote two macros, %LOF_LoOP and %LOF_LoOP_PLOT, to implement local outlier identification for subsequent treatment.

The conference paper and the %LOF_LoOP macros are included as attachments.