Why SAS?

• Author: Francesca Torti - Marco Riani - Anthony C. Atkinson - Domenico Perrotta - Aldo Corbellini
• Profession: European Commission, Joint Research Centre (JRC) - University of Parma, Italy - London School of Economics, UK - European Commission, Joint Research Centre (JRC) - University of Parma, Italy
• Pages: 12-15

0 100 200 300 400 5 00

1.5

2

2.5

3

3.5

4

CALL FSR("lib.loyalty", {’x1’ ’x2’ ’x3’}, "y", "CLASSIFY mdrplot ")

transform_original_data = 0.4 ;

100 200 300 400 500

−6

−4

−2

0

2

CALL FSR("lib.loyalty", {’x1’ ’x2’ ’x3’}, "y", "CLASSIFY RESFWDPLOT ")

transform_original_data = 0.4 ;

Figure 2: Loyalty card data: monitoring plots on for transformed data with λ= 0.4. The top panel shows

the absolute values of minimum deletion residuals among observations not in the subset; the last part of the

curve, corresponding to the 18 identified outliers, is automatically highlighted in red (in the on-line .pdf version).

The bottom panel shows the scaled residuals, with the trajectories corresponding to the 18 detected outliers

automatically represented in red (in the on-line .pdf version). The box under each panel contains the SAS code

used to generate the plot.

7. Why SAS?

The statistical community currently has three main environments for program development, which target

rather dif‌ferent market segments.

The R environment1is the most popular among statisticians and of‌fers many packages for robust statis-

tics, for example rrcov for multivariate analysis (Todorov and Filzmoser, 2009) and robustbase for re-

gression, univariate and multivariate analysis (Rousseeuw et al., 2009). Recently Riani et al. (2017) have

developed FSDAr for regression analysis.

Engineers and practitioners in physics, geology, transport, bioinformatics, vision and other f‌ields usually

prefer MATLAB, but f‌ind in the default distribution only a few robust tools, such as the Minimum Co-

variance Determinant (MCD, Hubert and Debruyne, 2010, introduced in the 2016 release through function

robustcov) and robust regression computed via iteratively re-weighted least squares (functions robustfit

and fitlm). Many more robust procedures are provided by two open toolboxes: Library for Robust Analysis

1R is available from the CRAN website: https://cran.r-project.org/

12