Page 212 - Special Topic Session (STS) - Volume 3
P. 212

STS540 J.-C. Malela-Majika et al.
                            Table 1. Recent runs-rules contributions to the literature
                                         NSS       Normal
                    Authors     Year      or          or                 Detail
                                         SS      non-normal
                                                               Investigated NSS and
                                                               various SS runs-rules
                  Shongwe &     2016  NSS & SS     Normal      and synthetic 
                                                                             ̅
                  Graham                                       schemes for normally
                                                               distributed data


                  Malela-                                      Proposed NSS SRR2-of-(h+1)  and
                  Majika,                                      IRR2-of-(h+1)   schemes for
                                                                         ̅
                  Kanyama &  2017        NSS     Non-normal  non-normal data using the
                  Rapoo                                        Burr-type XII distribution

                  Malela-                                      Proposed SS SRR2-of-(h+1)
                  Majika,                                      and IRR2-of-(h+1) Shewhart-
                  Malandala     2019      SS     Non-normal    type  schemes for non-
                                                                    ̅
                  & Graham                                     normal data


                      To differentiate between non-side-sensitive (NSS) and side-sensitive (SS)
                   chart, let’s start with the latter. The SS SRR2-of-(h+1) schemes signal when two
                   (out of ℎ  +  1) consecutive plotting statistics plot in Zone 1 (or Zone 3), which
                   are  separated  by  at  most  ℎ  −  1  plotting  statistics  that  plot  in  Zone  2,
                   whereas for NSS schemes signal whether some (or both) plotting statistics
                   fall in Zone 1 and others (or both) in Zone 3 (see Figure 1). The probabilities
                   of the plotting statistic plotting in Zones 1, 2 and 3, respectively, can easily
                   be computed, but the details are omitted here for conciseness. Only focusing
                   on SS from this point forward, the IRR2-of-(h+1) schemes signal when either a
                   single plotting statistics plots in Zone A (or Zone E) or when 2 (out of ℎ  +  1)
                   consecutive plotting statistics plot in Zone B (or Zone D), which are separated
                   by at most ℎ  −  1 plotting statistics that plot in Zone C (see Figure 2). The
                   probabilities of a plotting statistic plotting in Zones A, B, C, D and E, denoted,
                   respectively, can easily be computed, but the details are omitted here for
                   conciseness.
                      From Table 1 it can be seen that the contribution of this paper is that side-
                   sensitive  SRR2-of-(h+1)  and  IRR2-of-(h+1)  Shewhart-type  schemes  for  non-
                                                                       ̅
                   normal data are proposed (when the normality assumption fails to hold)
                   as  alternative  to  the  traditional  (parametric)  side-sensitive  SRR2-of-(h+1)  and
                   IRR2-of-(h+1) Shewhart-type  schemes.
                                             ̅
                      The remainder of this paper is structured as follows. Section 2 the design
                   of the proposed schemes under the Burr-type XII distribution is given. In
                                                                     201 |I S I   W S C   2 0 1 9
   207   208   209   210   211   212   213   214   215   216   217