See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/323258596

Robust dual-CUSUM control charts for contaminated processes

Article  in  Communication in Statistics- Simulation and Computation · February 2018

DOI: 10.1080/03610918.2018.1435804

CITATIONS READS

0 83

3 authors, including:

Nasir Abbas
King Fahd University of Petroleum and Minerals
39 PUBLICATIONS   445 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Maxwell project View project

An adaptive EWMA control chart based on CUSUM accumulation error. View project

All content following this page was uploaded by Nasir Abbas on 19 February 2018.

The user has requested enhancement of the downloaded file.

 Communications in Statistics - Simulation and
Computation

ISSN: 0361-0918 (Print) 1532-4141 (Online) Journal homepage: http://www.tandfonline.com/loi/lssp20

Robust dual-CUSUM control charts for

contaminated processes

Ishaq Adeyanju Raji, Muhammad Riaz & Nasir Abbas

To cite this article: Ishaq Adeyanju Raji, Muhammad Riaz & Nasir Abbas (2018): Robust dual-
CUSUM control charts for contaminated processes, Communications in Statistics - Simulation and
Computation, DOI: 10.1080/03610918.2018.1435804

To link to this article: https://doi.org/10.1080/03610918.2018.1435804

Published online: 18 Feb 2018.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=lssp20
COMMUNICATIONS IN STATISTICS—SIMULATION AND COMPUTATION
, VOL. , NO. , –
®
https://doi.org/./..

Robust dual-CUSUM control charts for contaminated processes

Ishaq Adeyanju Rajia,b , Muhammad Riazb , and Nasir Abbasb
a
Department of Mathematics, Universiti Teknologi Malaysia, Johor Bahru, Malaysia; b Department of
Mathematics and Statistics, King Fahd University of Petroleum and Minerals, Dhahran, Saudi Arabia

ABSTRACT ARTICLE HISTORY

In this article, we improve the efficiency of the Dual CUSUM chart (which Received  November 
combines the designs of two CUSUM structures to detect a range of Accepted  January 
shift) by focusing on its robustness, ability to resist some disturbances
KEYWORDS
in the process environment and violation of basic assumptions. We do Average run length;
that, by proposing some robust estimators for constructing the chart contamination; CUSUM;
for both contaminated and uncontaminated environments. The average hodges-lehmann; robustness
run length is used as the performance evaluation measure of the charts.
After comparing the performances of the proposed charts based on the
estimators, it is noticed that the tri-mean estimator out-performs oth-
ers in all ramifications. Next to it in performance is the Hodges-Lehmann
and midrange estimators. We substantiated the simulation results of the
study by applying the scheme on a real-life data set.

1. Introduction
Statistical Process Control (SPC) is a set of techniques used to control and monitor the spe-
cial causes of variation in manufacturing or service processes, since the natural variations
are inevitable part of the processes. These techniques are called the magnificent seven of SPC,
namely: cause and effect diagram, Pareto charts, histogram, flow charts, probability plots, scat-
ter diagram and control chart Montgomery (2012). It is truism that amongst these techniques,
control chart is the most efficient and widely used of all. Processes are said to be statistically
In Control (IC) with the presence of natural variations only, but they are considered Out of
Control (OC) due to the existence of unnatural variations. Control charts are classified, (with
respect to the magnitude of variation they are designed to monitor) into memory-less charts
and the memory type. The memory-less chart uses the present information of the process,
truncating the past information in constructing the chart statistic. This makes it less sensitive
to detect small and moderate shift but highly sensitive to large shifts in the process. While
the memory chart makes use of both present and past information of the process in con-
structing the chart statistics and by so doing makes it more sensitive to small and moderate
shifts in the process. An example of the memory-less chart is the Shewhart chart named after
Shewhart (1931). Examples of the memory chart are Cumulative Sum (CUSUM) by Page
(1954) and Exponential Weighted Moving Average (EWMA) by Robert (1959). Both memory
and memory-less charts are used to monitor and control spread and/or location parameters
in any process.

CONTACT Nasir Abbas nasirabbas@yahoo.com Department of Mathematics and Statistics, King Fahad University
of Petroleum and Minerals, Dhahran , Saudi Arabia.
Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/lssp.
©  Taylor & Francis Group, LLC
2 I. A. RAJI ET AL.

The importance and effectiveness of the memory charts have attracted the interest of many
researchers, leading to many modifications and improvements in the field. In the quest of
increasing the efficiency and performance of these charts, various type of charts are available
in literature. To refer a few of them; Estimating common-cause sigma proposed by Boyels
(1997), to control both large and small shifts with a single chart. Lucas and Crosier (1990)
introduced a Fast Initial Response (FIR) feature to the classical CUSUM in order to detect
small and moderate shifts faster than the ordinary CUSUM.
In the same vain, Zhao, Tsung, and Wang (2005) designed a dual CUSUM (DCUSUM)
scheme that combines two single-sided CUSUM statistics and monitors them simultaneously
to detect a range of small shifts in a process. This idea is an adjustment and improvement
to that pioneered by Lordens (1971) where a combined set of infinite number of classical
CUSUM statistics are plotted simultaneously against their respective limits. Lordens (1971)
proved the asymptotical optimality feature of this chart by claiming the smallest OC Average
Run Length (ARL) when the IC ARL tends to infinity. Unfortunately, there was no empirical
or simulation illustration in literature to buttress his claim.
The underlying assumptions of all the aforementioned charts are that the environment of
these processes are normal, uncontaminated and free from outliers. In a situation where any
of these assumptions is not met, the efficiency of the corresponding chart becomes question-
able. One of the ways to address this matter is to use some robust charts that can not only
resist the violation of the assumptions but also maintain the efficiency of their performances.
Series of publications are available in literature that incorporate robustness in the charts. Most
of these techniques are based on transforming the data into a non-parametric statistics and
then using the transformed statistics in the CUSUM schemes. (see, Ion and Klaassen (2006);
Li (2015) and Haq (2018)). To mention a few: a robust control chart for estimating disper-
sion is designed by Tatum (1997); Moustafa and Mokhtar (1999) constructed another robust
control chart using the Hodges-Lehman and Shamos-Bickel-Lehman estimators for location
and dispersion parameters, simultaneously. Lately, Riaz, Abbas, and Does (2011) presented a
comprehensive study on control charts whose ideal assumptions are violated. Their properties
and effects on their performances were discussed in details.
On the contrary, Yang, Pai, and Wang (2010) proposed a robust CUSUM control chart
using the sample median as the chart statistic where the observed data is not transformed
to a non-parametric statistic. Muzaffar, Deng, and Rashid (2016) also designed a non-
parametric optimal service pricing. In extension, Nazir et al. (2013) designed a set of
robust CUSUM control charts with some robust location estimators and compared their
effectiveness and robustness for phase II monitoring. Zwetsloot, Schoonhoven, and Does
(2016) also designed a robust point location estimators for the EWMA control chart. (see
Abbas et al. (2017) and Riaz, Khaliq, and Gul (2017)). Nazir et al. (2016) provided a
thorough comparison of Shewhart, CUSUM and EWMA charts under different types of
contaminations. In this article, we propose a set of robust dual CUSUM control chart-
ing schemes with five different location estimators for monitoring phase II samples. We
evaluate these charts based on their run length distribution (ARL and Standard Devia-
tion Run Length (SDRL)). The performances are examined in different environments i.e.
the uncontaminated, the location contaminated and the dispersion contaminated normal
environments.
The remainder of this article is organized as follows: the next section, we discuss the clas-
sical CUSUM control chart versus the construction and design of the DCUSUM control
chart; Section 3 is a detailed description of the proposed estimators and their correspond-
ing DCUSUM chart; the performance evaluation of these proposed charts are presented in
 COMMUNICATIONS IN STATISTICS—SIMULATION AND COMPUTATION ® 3

Table A. Nomenclature.
S/N Notations/Symbols Meaning

 ARL Average Run Length

 ARL0 In control Average Run Length
 ARL1 Out-of- Control Average Run Length
 CL Centre Line
 CUSUM Cumulative Sum
 DCUSUM Dual Cumulative Sum
 DEWMA Double Exponentially Weighted Moving Average
 EWMA Exponentially Weighted Moving Average
 FIR First Initial Response
 HL Hodges- Lehmann
 IC ARL In control Average Run Length
 LCL Lower Control Limit
 LCNE Location Contaminated Normal Environment
 MR Mid-Range
 OC ARL Out of Control Average Run Length
 SDRL Standard Deviation Run Length
 TM Tri-Mean
 UCL Upper Control Limit
 UNE Uncontaminated Normal Environment
 VCNE Variance Contaminated Normal Environment

Section 4; Section 5 buttresses all the proposals with application on a real life dataset; finally,
summary and conclusions appear in Section 6. Also, Table A provides the summary of all the
symbols and notations that are to be used in the rest of this article.

2. Classical cusum and dual cusum control charts

In this section, we give a detailed explanation of the classical CUSUM and its counterpart.
Subsection 2.1 entails the explanation of the classical CUSUM, while subsection 2.2 explains
the formation of the Dual CUSUM. In subsection 2.3, comes the selection of the dual CUSUM
parameters.

2.1. The classical cusum chart

The classical CUSUM chart pioneered by Page (1954) is one of the most popular control chart
for monitoring small and moderate shifts in processes. This chart, like other counterparts,
can be used for individual data and subgroups (i.e. sample size n > 1). The classical CUSUM
control chart comprises of two statistics Ci+ and Ci− which are plotted against a single limit
interval H. the statistic is defined as:
 
Ci+ = max 0, (θ̂i − μθˆ ) − K + Ci−1
+
(1)

where i is the subgroup number, θˆi is the estimator of the study variable under consideration.
μθˆ = μ0 is the target mean of the study variable, C0+ = C0− = 0 and K = kσθˆ is the reference
value of the CUSUM scheme. The next task is to plot each of these two statistics against H =
hσθˆ . If Ci+ > H for any value of i we conclude that the process has shifted upward and if
Ci− > H at any point i, the process mean is declared to have moved downwards. When n = 1,
θˆi is replaced by the corresponding individual observation Xi . The parameters that define this
chart are K and H, which are to be selected using a predetermined IC Average Run Length
(ARL0 ) of interest. Therefore, a careful selection of these parameters should be put in place
in order not to affect the performance of the chart. Researchers have published a number of
4 I. A. RAJI ET AL.

articles where the selection of these parameters and their effect on the chart were studied in
details.

2.2. Construction of dual-Cusum control chart

Assume the variable of interest X of a particular process happens to follow a normal dis-
tribution i.e. X ∼ N(μ0 , σ0 ). The ideal and appropriate CUSUM chart for this process is a
one-sided CUSUM chart that uses only Eq. 1 and plots it against an interval limit H to detect
upward shifts in the process mean. The process is expected to always be in an IC state, except
the mean of the process shifts to say μ1 = μ0 + δ, that warrants the statistic to be greater than
H so that the process is declared OC.
Similarly, extending this CUSUM chart to DCUSUM, we have two single-sided CUSUM
statistics measured and monitored simultaneously as defined below:
 
C1i+ = max 0, (θ̂i − μθˆ ) − K1 + C1i−1
+
(2)
 
C2i+ = max 0, (θ̂i − μθˆ ) − K2 + C2i−1
+
(3)

The two statistics in Eqs (2) and (3) are characterized by different parameters K1 & H1 for
C1i+and K2 & H2 for C2i+ . If C1i+ > H1 and/or C2i+ > H2 at any time i the chart sends a signal
and concludes that process is OC at that time. C10 + +
= C20 = 0. θˆi is a location estimator for ith
subgroup of size n. It is important to note that this chart can be used to monitor spread as
well.

2.3. Selection of the chart parameters

In a classical CUSUM charts, the parameters involved are K and H which are selected against
a specified ARL0 . In the same manner, we choose K1 & H1 for C1+ and K2 & H2 for C2+ . Zhao,
Tsung, and Wang (2005) discussed the conditions and criteria for selecting these parameters
without keeping any of the statistic C1+ and C2+ redundant or ineffective. They argued that for
the DCUSUM chart to be effective, following criteria must be met and not violated in selecting
its parameters.
K1 < K2
H1 > H2 + K2 − K1 (4)
K2 /K1 ∼
= H1 /H2
Zhao, Tsung, and Wang (2005) intensively justified these conditions logically and alge-
braically. Let K1 < K2 so that for any i ∈ N, we have C1i+ > C2i+ which ensures C2i+ is not
redundant. Also there exist a point i, where C2i+ > H2 which implies C1i+ > C2i+ > H2 >
0. Recall (from DCUSUM equation) C1i+ − C2i+ = K2 − K1 + C1i−1 + +
− C2i−1 , also, since C1i+ >
+ + + + +
C2i , this implies C1i−1 − C2i−1 > 0, it then follows that C1i − C2i ≥ K2 − K1 for any i, and
whenever C2i+ triggers an out of control signal (i.e. C2i+ > H2 ), we get that C1i+ > H2 +
K2 − K1 . Now suppose H1 ≤ H2 + K2 − K1 this insinuates that whenever C2i+ triggers an
out-of-control signal, C1i+ has already sent a signal as well. Then C2i+ is actively working
uselessly.
Following the justification of the first two conditions, we note that there are infinitely
possible combinations of K1 , K2 & H1 , H2 such that these conditions are met. The question is
which of them is to be chosen? This question takes us back to the asymptotic optimal feature
 ®
COMMUNICATIONS IN STATISTICS—SIMULATION AND COMPUTATION 5

of Lordens’ control chart, which simultaneously monitor an infinite number of classical

CUSUM statistic defined below:
 
Gnm = sup kSnm − (n − m) k2 . (5)
k(a/2,b/2)

He (Lordens) proved that this chart has the lowest OC ARL when ARL0 tends to infinity,
meeting the restriction that the multiplication of the parameters of all individual CUSUM
charts must be equal i.e. k1 h1 = k2 h2 = k3 h3 = . . . ki hi = h. The third condition above fol-
lows from applying this restriction for the DCUSUM. Taking all these three axioms into con-
sideration, we carefully selected the parameters k1 and k2 from a Monte Carlo simulation to
be 0.12 and 0.5 respectively.

3. Description of the proposed charts

An estimator is defined as a function used in calculating an estimate of an estimand based on
observed dataset. The estimand (the quantity of interest) could be a location or spread param-
eter. An estimator is said to be robust, if it maintains its performance for a range of dataset
drawn from different probability distributions. Hence robust estimators seek to provide resis-
tance to contaminations, outliers and violation of the assumptions of the distribution under
consideration which duly affect the method or process.
In this article, we study the performance of five robust estimators with the DCUSUM chart
for the phase II analysis, namely: mean, median, mid-range, Hodges-Lehmann and Tri-mean.
These estimators have been studied for the phase I analysis by Schoonhoven et al. (2011), and
studied with classical CUSUM chart for the phase II analysis by Nazir et al. (2013).
Furthermore, these set of estimators have been proved to be robust against contaminations
and outliers by other authors (see Abbas and Fried (2017)). One of the reasons of choosing
these estimators amongst other estimators is they are parametric estimators. On the other
hand, in order to achieve a fair ground of comparison, Zhao, Tsung, and Wang (2005) used
the mean estimator for the DCUSUM chart, while we combine the other four parametric
estimators: median, midrange, trimean and Hodges-Lehmann estimators, as we use the mean
as the benchmark of comparison.
Recall the DCUSUM chart defined in Eqs 2 and 3, the estimator θˆ estimating the location
parameter θ can be replaced with any of the following estimators. Based on a random sample
of size n with observation {x1 , x2 , . . . xn } we define:
Meanx̄: Is the summation of all the observation in a sample divided by the sample size.
n x
i.e. x̄ = i=1n
i
. The sample mean is not a robust estimator, though it’s a useful and common
measure of central tendency, especially for processes that follow a normal distribution. It is
included in the study as a base of comparison between others.
Median x̃: Unlike the mean, x̃ is a robust estimator of location. It has a breakdown point of
50%, whereas the mean has a breakdown point of 0%. It is the numerical value that separates
the higher half of a sample from the lower half. For odd number of sample, median is the
middle order statistic, and the average of the two middle-order statistic for an even number.
Mid-range: It is the average of the largest and lowest observation in a sample. It can also
be defined as the mid-point of range. Due to its composition, the two extreme values in the
sample, makes it very sensitive to outliers. Furthermore, it is also a non-robust estimator as it
ignores all other elements of the sample in its computation, having a breakdown point of 0%.
Hodges – Lehmann: This estimator was proposed by Hodges and Lehmann (1963). It is
also called Hodges-Lehmann-Sen estimator. It is the median of the pairwise averages of all the
6 I. A. RAJI ET AL.

observation in a sample. HL = median{(xi + x j )/2, 1 ≤ i < j ≤ n}. It is a robust estimator

with a breakdown point of 29% and at least 95% of relative efficiency compared to mean with
the presence of contaminations and outliers in the probability distribution under considera-
tion. Especially when the probability distribution is normal, the relative efficiency is always
above unity.
Tri-Mean: the tri-mean, also referred to as Turkey tri- mean, is defined as the weighted
average of the sample median and its two quantiles. TM = (Q1 + 2Q2 + Q3 )/4. It is also
equivalent to the average of the median and the mid-hinge TM = 12 [(Q2 + (Q1 + Q3 )/2)]. It
is a robust estimator of location that combines the median’s emphasis on center values with
the mid-hinge’s attention to the extremes. It has a breakdown point of about 25% and 88%
efficiency relatively to the sample mean.
Therefore, these proposed estimators of location are used in place of θˆ in the DCUSUM
chart (given in Eqs 2 and 3). This in turn constitutes five different DCUSUM charts. Each of
them is parameterized by K1 & K2 and H1 & H2 . The selection of these parameters is explained
in the next section.

4. Algorithms and performance evaluation of the proposed charts

This section explains the general algorithm of the control chart alongside the perfor-
mance of the DCUSUM charts and depicts their comparisons. We consider the standard
normal distribution as the parent environment of the process. The general basis of mea-
suring control charts lies solely on the run length distribution of the charts. Run length
is therefore the sequence/series of points plotted on a chart until a point gets out of
the control limit then a signal is raised. Amongst all the performance measures of con-
trol charts, ARL and SDRL are constructed directly from the run length variable, while
other measures such as EQL, SEQL, RARL and so on, are all functions of ARL. This
shows how important ARL and SDRL are in measuring the performance of any control
chart
Furthermore, ARL and SDRL are the most commonly used performance measure in con-
trol charts. ARL is defined as the average number of sample points until the first out of control
(OC) signal occurs. There are two types of ARL. The in control ARL (IC ARL) is the average
run length of a process, when there is no shift in the process mean (target). This should be
maximized such that the false alarm is delayed as much as possible when the process is IC. The
out of control ARL (OC ARL) is the average run length when the process mean has shifted
from its target, which should be minimized so that the signal is given as early as possible when
the process goes out of control.
On the other hand, the standard deviation run length (SDRL) is a measure of the spread in
the run lengths around their centre. It is a very useful quantity to ensure the stability of run
length behaviour. Therefore, to evaluate the performance of this control charting schemes, we
adopt the Monte Carlo Simulation (whose algorithm is explained shortly) to obtain the charts’
average run lengths (ARL). The IC ARL is denoted by ARL0 and the OC ARL is denoted
by ARL1 . We also report the standard deviation of the run length (SDRL) to have a better
knowledge of the run length distribution. The algorithm we employed for the proposed chart
is thus:
(1) Generate a subgroup of 105 random numbers with a fixed sample size n each from
a standard normal distribution. Then calculate each of the five estimators aforemen-
tioned (see 3.1 – 3.5), and independently used them in computing the DCUSUM statis-
tics C1i+ and C2i+ given in Eqs (2 and 3).
 ®
COMMUNICATIONS IN STATISTICS—SIMULATION AND COMPUTATION 7

(2) Plot the C1i+ and C2i+ against their control limits as specified in Eqs (2 and 3) follow-
ing the axioms in the inequalities (4). Whenever C1i+ and C2i+ are greater than any
of their respective limits, the process is out of control (OC). Terminate the program
and record the point i which the program went OC. Apply this to each of the five
estimators.
(3) Repeat steps (1 and 2) for 105 runs. Then find the averages standard deviations of all
points i recorded for which the process sent a signal. The averages are termed ARL
and the standard deviations are SDRL. The shifts we considered in this study lie in the
range of [0, 1] both inclusive.
Adopting the algorithm described, we derive the control limit coefficient H1 and H2 for
each of the five estimators under consideration for sample size n = 5, k1 = 0.12, k2 = 0.5
and a specified ARL0 = 370. For our study purposes our derived H1 and H2 is shown in the
table below:

Estimators H1 H2

Mean . .

Median . .
Midrange . .
Hodges Lehmann . .
TriMean . .

To accomplish one of the aims of this article, which is examining the robustness of the
DCUSUM charts, we introduce some location and dispersion contaminations to the standard
normal environment and study their behaviours. Therefore, the simulation results we present
are in three folds, the uncontaminated normal environment, the location contaminated and
variance contaminated environments. And each of the contamination is of two levels, the 5%
and 1% contaminations.

4.1. Uncontaminated normal environment

An uncontaminated Normal environment is one which all observations are drawn from
N(μ0 , σ02 ) The ARL’s and SDRL’s of DCUSUM chart using the different five estimators
(earlier mentioned) under an uncontaminated standard normal distribution are shown in
Table 1.
In Table 1, we can see that all the five estimators have an in-control (when δ = 0) ARL
clustering around 370 which is the target. But in other cases when δ = 0, the story changes.
Though they are all monotonically decreasing as the shift δ increases. It should be noted that
the performance of a control chart is evaluated based on how fast it sends an out-of-control
signal when the process under monitoring is actually out of control. This is explained in the
spirit of run length by the lower values of ARL1 s in an OC process. Of course, the sample mean
is the best estimator under an uncontaminated normal environment. The Hodges-Lehmann
estimator, therefore can be compared to the sample mean, as its ARL1 s are almost the same
in all cases of shift. Next to these two in performance is the Tri-mean estimator, while others
are out rightly larger than the sample mean which is the yardstick of comparison. Further-
more, the SDRL behaviour for all the estimators is positively skewed by increment in the shift.
Like the ARL, the lower the SDRL, the better the chart performance in detecting small shifts.
(cf. Table 1 and Figure 1).
8 I. A. RAJI ET AL.

Table . ARL and SDRL values for the DCUSUM with k1 = 0.12, k2 = 0.50, and ARL0 = 370.
S hift = . . . . . . . . . . . .

Mean
ARL . . . . . . . . . . . .
SDRL . . . . . . . . . . . .
Median
ARL . . . . . . . . . . . .
SDRL . . . . . . . . . . . .
MR
ARL . . . . . . . . . . . .
SDRL . . . . . . . . . . . .
TM
ARL . . . . . . . . . . . .
SDRL . . . . . . . . . . . .
HL
ARL . . . . . . . . . . . .
SDRL . . . . . . . . . . . .

4.2. Location - contaminated normal environment (LCNE)

A location Contaminated Normal environment is a model/distribution whose location
parameter has been disturbed to a certain percent. Say, A random variable X, is said to be

Figure . The ARL1 curves and corresponding ARL0 values for the normal environment, LCNE and VCNE.
 COMMUNICATIONS IN STATISTICS—SIMULATION AND COMPUTATION® 9

Table . ARL and SDRL values DCUSUM % LCNE k1 = 0.12, k2 = 0.50 ARL0 = 370.
S hift = . . . . . . . . . . . .

Mean
ARL . . . . . . . . . . . .
SDRL . . . . . . . . . . . .
Median
ARL . . . . . . . . . . . .
SDRL . . . . . . . . . . . .
MR
ARL . . . . . . . . . . . .
SDRL . . . . . . . . . . . .
TM
ARL . . . . . . . . . . . .
SDRL . . . . . . . . . . . .
HL
ARL . . . . . . . . . . . .
SDRL . . . . . . . . . . . .

drawn from a LCNE, if it has (ϕ)100% observations from N(μ0 , σ02 ) and (1 − ϕ)100% obser-
vations from N(μ0 + ωσ0 , σ02 ) where −∞ < ω < ∞. In this section, we study the run length
distribution of the DCUSUM charts with ϕ = 0.05 & 0.01 and ω = 4, with the proposed esti-
mators. In Table 2 for ϕ = 0.05 and Table 3 for ϕ = 0.01 respectively, it is obvious that all
the five estimators are not adequate for the target ARL0 as they are drastically lower than
ARL0 = 370. Though, despite the poor performances, we can still compare the estimators
based on their resistance and robustness to the disturbances.
In both cases, when median turns to have the highest ARL0 , followed by tri-mean and
Hodges-Lehmann. The remaining estimators are therefore not comparable with the first three
due to their low performances. It should be noted that, the substantial drop in the ARL0 values
of these estimators is because they are measures of location and the disturbance is also on
the location parameter. This can be justified by the increase in the values of ARL 0 s for all
the estimators, as the level of disturbance ϕ decreases. In Table 3, when ϕ = 0.01, the ARL 0 s
values of the estimators are although not comparable with that of the uncontaminated normal
environment, but substantially greater compared to when ϕ = 0.05. This poor performance
in the ARL0 , thwarted the aim of analysing the ARL 1 s results and comparing the estimators
based on these values (cf. Tables 2-3 and Figure 1).

Table . ARL and SDRL values DCUSUM % LCNE k1 = 0.12, k2 = 0.50 ARL0 = 370.
S hift = . . . . . . . . . . . .

Mean
ARL . . . . . . . . . . . .
SDRL . . . . . . . . . . . .
Median
ARL . . . . . . . . . . . .
SDRL . . . . . . . . . . . .
MR
ARL . . . . . . . . . . . .
SDRL . . . . . . . . . . . .
TM
ARL . . . . . . . . . . . .
SDRL . . . . . . . . . . . .
HL
ARL . . . . . . . . . . . .
SDRL . . . . . . . . . . . .
10 I. A. RAJI ET AL.

Table . ARL and SDRL values DCUSUM % VCNE k1 = 0.12, k2 = 0.50 ARL0 = 370.
S hift = . . . . . . . . . . . .

Mean
ARL . . . . . . . . . . . .
SDRL . . . . . . . . . . . .
Median
ARL . . . . . . . . . . . .
SDRL . . . . . . . . . . . .
MR
ARL . . . . . . . . . . . .
SDRL . . . . . . . . . . . .
TM
ARL . . . . . . . . . . . .
SDRL . . . . . . . . . . . .
HL
ARL . . . . . . . . . . . .
SDRL . . . . . . . . . . . .

4.3. Variance - contaminated normal environment (VCNE)

The VCNE like the LCNE, is a distribution whose spread parameter has been disturbed. A
random variable say X, is said to be drawn from a VCNE, if it has (ϕ)100% observations from
N(μ0 , σ02 ) and (1 − ϕ)100% observations from N(μ0 , τ σ02 ) where −∞ < τ < ∞. Similarly,
we study the performances of the proposed estimators with their corresponding charts under
this environment. We examined when ϕ = 0.05 & 0.01 and τ = 9. Table 4 and 5 show the
ARL and SDRL values for both IC and OC cases.
Generally, all the estimators under the variance-contaminated environment performed
better compared to the location contaminated environment due to the fact that the estimators
are estimating location parameter. Nonetheless, under ARL0 , the sample median out performs
other estimators, as it has the closest value to the target, especially when the percent of con-
tamination is low. Next to the median, is the TM and HL, which have comparable values to
the median. (cf. Tables 4-5 and Figure 1)

5. Application of the schemes on a real life dataset

Claims without proofs are mere expressions they say. In this section, we have substantiate
the efficiency of these charts by a practical demonstration with a real life data set. The data

Table . ARL and SDRL values DCUSUM % VCNE k1 = 0.12, k2 = 0.50 ARL0 = 370.
S hift = . . . . . . . . . . . .

Mean
ARL . . . . . . . . . . . .
SDRL . . . . . . . . . . . .
Median
ARL . . . . . . . . . . . .
SDRL . . . . . . . . . . . .
MR
ARL . . . . . . . . . . . .
SDRL . . . . . . . . . . . .
TM
ARL . . . . . . . . . . . .
SDRL . . . . . . . . . . . .
HL
ARL . . . . . . . . . . . .
SDRL . . . . . . . . . . . .
 COMMUNICATIONS IN STATISTICS—SIMULATION AND COMPUTATION ® 11

Figure . The control charts of DCUSUM scheme for the five estimators under normal environment.

set which consists of 30 sample points with the sample size n = 5, is the waiting time in min-
utes of patients for a colonoscopy procedure in a regional health center. Jones-Farmer, Jordan,
and Champ (2009). The original data set without contamination is presented in Figure 2(a),
while the subsequent Figure 2(b–f) show the DCUSUM control chart of respective charts.
For all the estimators, their respective charts behave quite well under the uncontaminated
normal environment as we can see all the points of the charts falling within the control
limits.
In the location contaminated environments, comes the survival of the fittest rule. As we can
see from Figures 3(b) and 3(d) that both mean and midrange could not withstand the location
contamination applied to the original data set. At sample point 14, the midrange chart went
out of its limit till point 17, which makes it the least robust of all the estimator. Next to it in
bad performance, is the mean. It becomes OC process at sample point 15 and maintains its
limit back immediately at point 16.
Out of the five estimators, the median chart proves to be the best, as we can see from
the chart, followed by trimean and Hodges-Lehmann estimators hierarchically. The real-life
example as just buttressed the simulation result produced in that regard.
The same scenario occurs under the variance contaminated environment sample, the
median retains its position as the best of all, followed by trimean and Hodges-Lehmann
chart respectively. Though the performance of the median and mean charts under the vari-
ance contamination seems to be better than that of location contamination, but relatively,
12 I. A. RAJI ET AL.

Figure . The control charts of DCUSUM scheme for the five estimators under LCNE.

the two charts are still in an OC state at some points, as it is obvious in their respective
charts (cf. Figure 4).

6. Summary and conclusion

Control charts are the widely used tool of SPC for monitoring manufacturing and service pro-
cesses. A memory type of this chart is the classical CUSUM chart that is efficient in detecting
small and moderate shifts in any process. The one-sided DCUSUM chart was proved to have
better performance compared to the classical CUSUM in terms of the ARL & SDRL, while the
former was designed to detect a range of varying shifts in a process, the latter was designed to
monitor a specified shift in a process. In this study, we establish that robust DCUSUM charts
are preferred to the ordinary DCUSUM charts as robust classical CUSUM were preferred to
classical CUSUM. Adopting some robust statistics in the presence of violations of normality
assumptions under normal environments such as contaminations and outliers, makes the cor-
responding charts maintain their efficient and robustness. Amongst the five location estima-
tors used in designing the DCUSUM charts, with the results shown above, the mean estimator
happens to be the most efficient of all under non- contaminated normal environments, while
the TM and HL estimators are good comparisons to the mean. In the presence of location con-
tamination, the performance of all the estimators are quite unsatisfactory, which is expected,
 COMMUNICATIONS IN STATISTICS—SIMULATION AND COMPUTATION® 13

Figure . The control charts of DCUSUM scheme for the five estimators under VCNE.

because the estimators are estimating location parameters. Despite this worst performance,
the TM and HL estimators’ charts outperform others, based on ARL0 and ARL1 examination.
Though the ARL0 of the median seems to be the best, but the median chart performance is
slow compared to TM and HL performances in detecting small shifts in the process which is of
paramount. The performance of these charts under dispersion contamination environments
depicts the efficiency and robustness of the TM and HL estimators better.
In general, this study substantially proves that the DCUSUM chart is preferred to the
ordinary CUSUM, and TM estimator is the best amongst the five estimators under all
circumstances.

Acknowledgments
The authors Muhammad Riaz and Nasir Abbas would like to acknowledge the support provided by
Deanship of Scientific Research (DSR) at King Fahd University of Petroleum and Minerals for funding
this work through project No. IN161037.

Funding
King Fahd University of Petroleum and Minerals (IN161037).
 14 I. A. RAJI ET AL.

References
Abbas, S., and R. Fried. 2017. Control charts for the mean based on robust two-sample tests. Journal of
Statistical Computation and Simulation 87 (1):138–55. doi:10.1080/00949655.2016.1194839.
Abbas, T., S. Ahmad, M. Riaz, and Z. Qian. 2017. A bayesian way of monitoring the linear pro-
files using CUSUM control charts. Communications in Statistics-Simulation and Computation.
doi:10.1080/03610918.2017.1375520.
Boyles, R. A. 1997. Estimating common-cause sigma in the presence of special causes. Journal of Quality
Technology 29 (4):381–95.
Haq, A. 2018. A new nonparametric synthetic EWMA control chart for monitoring process mean. Com-
munications in Statistics-Simulation and Computation. doi:10.1080/03610918.2017.1422750.
Hodges, J. L. Jr., and E. L. Lehmann. 1963. Estimates of location based on rank tests. Annals of Mathe-
matical Statistics 34:598–611. doi:10.1214/aoms/1177704172.
Ion, R. A., and C. A. J. Klaassen. 2006. Nonparametric shewhart control charts. Journal of Nonparametric
Statistics 17 (8):971–88. doi:10.1080/10485250500393271.
Jones-Farmer, L. A., V. Jordan, and C. W. Champ. 2009. Distribution-free phase I charts for subgroup
location. Journal of Quality Technology 41 (3):304–16. doi:10.1080/00224065.2009.11917784.
Li, J. 2015. Nonparametric multivariate statistical process control: A hypothesis testing-based approach.
Journal of Nonparametric Statistics 27 (3):384–400. doi:10.1080/10485252.2015.1062889.
Lorden, G. 1971. Procedures for reacting to a change in distribution. Annals of Mathematical Statistics
42 (6):1897–908. doi:10.1214/aoms/1177693055.
Lucas, J. M., and M. S. Saccucci. 1990. Exponentially weighted moving average control schemes: Prop-
erties and enhancements. Technometrics 32 (1):1–12. doi:10.1080/00401706.1990.10484583.
Montgomery, D. C. 2012. Introduction to statistical quality control. (7th ed.), New York: Wiley.
Moustafa, B. A., and A. Mokhtar. 1999. New robust statistical process control chart for location. Quality
Engineering 12 (2):149–59. doi:10.1080/08982119908962572.
Muzaffar, A., S. Deng, and A. Rashid. 2016. Nonparametric optimal service pricing: A
simulation study. Quality Technology and Quatitative Management 14 (2):142–55.
doi:10.1080/16843703.2016.1208487.
Nazir, H. Z., N. Abbas, M. Riaz, and R. J. M. M. Does. 2016. A comparative study of memory-type
control charts under normal, contaminated and non-normal environments. Quality and Reliability
Engineering International 32 (4):1347–56. doi:10.1002/qre.1835.
Nazir, H. Z., M. Riaz, R. J. M. M. Does, and N. Abbas. 2013. Robust CUSUM control charting. Quality
Engineering 25 (3):211–24. doi:10.1080/08982112.2013.769057.
Page, E. S. 1954. Continuous inspection schemes. Biometrika 41:100–15. doi:10.1093/biomet/41.1-
2.100.
Riaz, M., N. Abbas, and R. J. M. M. Does. 2011. Improving the performance of CUSUM charts. Quality
and Reliability Engineering International 27 (4):415–24. doi:10.1002/qre.1124.
Riaz, M., Q. U. A. Khaliq, and S. Gul. 2017. Mixed tukey EWMA-CUSUM control chart
and its applications. Quality Technology & Quantitative Management 14 (4):1–34.
doi:10.1080/16843703.2017.1304034.
Roberts, S. W. 1959. Control chart tests based on geometric moving averages. Technometrics 1 (3):
239–50. doi:10.1080/00401706.1959.10489860.
Schoonhoven, M., H. Z. Nazir, M. Riaz, and R. J. M. M. Does. 2011. Robust location estimators for the
X control chart. Journal of Quality Technology 43:363–79. doi:10.1080/00224065.2011.11917870.
Shewhart, W. A. 1931. Economic control of quality of manufactured product. New York, NY: D. Van
Nostrand Company.
Tatum, L. G. 1997. Robust estimation of the process standard deviation for control charts. Technometrics
39:127–41. doi:10.1080/00401706.1997.10485078.
Yang, Y. L., S. Pai, and Y. Wang. 2010. A novel CUSUM median control chart. In International multi-
conference of engineers and computer scientists proceedings, Hong Kong, March 17–19, 2010.
Zhao, Y., F. Tsung, and Z. Wang. 2005. Dual CUSUM control schemes for detecting a range of mean
shifts. IIE Transactions 37 (11):1047–57. doi:10.1080/07408170500232321.
Zwetsloot, I. M., M. Schoonhoven, and R. J. M. M. Does. 2016. Robust point location estima-
tors for the EWMA control chart. Quality Technology & Quantitative Management 13 (1):29–38.
doi:10.1080/16843703.2016.1139845.

View publication stats