Computational Statistics & Data Analysis 51 (2007) 38623870

www.elsevier.com/locate/csda
The quality control chart for monitoring multivariate
autocorrelated processes
Jeffrey E. Jarrett
a,
, Xia Pan
b
a
University of Rhode Island, USA
b
University of Illinois at Springeld, USA
Received 13 July 2005; received in revised form 15 January 2006; accepted 17 January 2006
Available online 30 March 2006
Abstract
Previously, quality control and improvement researchers discussed multivariate control charts for independent processes and
univariate control charts for autocorrelated processes separately. We combine the two topics and propose vector autoregressive
(VAR) control charts for multivariate autocorrelated processes. In addition, we estimate AR(p) models instead of ARMA models for
the systematic cause of variation. We discuss the procedures to construct the VAR chart. We examine the effects of parameter shifts
and by example present procedures to show the feasibility of VAR control charts. We simulate the average run length to assess the
performance of the chart.
2006 Elsevier B.V. All rights reserved.
Keywords: Quality control and improvement; VAR control charts; Multivariate autocorrelated processes
1. Introduction and purpose
Shewhart (1931) developed statistical process control (SPC) techniques for the improvement and assurance of product
and service quality. The assumptions of these methods include serial independence of successive samples taken from
a process. Alwan and Roberts (1988) proposed modeling and ltering to monitor residuals by a one-step-forecast of
autocorrelated process possessing systematic patterns. English and Sastri (1990) applied Kalman lter to monitoring
univariate autoregressive (AR) (1) process. Wardell et al. (1994) derived the run length distribution of Shewhart
chart on residuals. Montgomery and Mastrangelo (1991), Lu and Reynolds (1999) discussed some SPC charts for
autocorrelated data. Runger and Willemain (1995) compared Shewhart charts based on several process transformations
to eliminate autocorrelation. Snoussi et al. (2005) indicated that residuals control charts provide much better shift
detection properties. Testik (2005) considered using an AR (1) process to modify control limits for residual based
charts.
Loredo et al. (2002) provided a method for monitoring autocorrelated processes based on Hawkins (1991, 1993)
regression adjustment. They examined the performance of residual based quality control charts in terms of the average
run length to observation based control charts via Monte Carlo simulations. Their results indicate the superiority of
residual based control in comparison with the observation based control charts when detecting a mean shift in short run

Corresponding author. Tel.: +1 401 8744169; fax: +1 401 8744312.

E-mail addresses: jejarrett@mail.uri.edu (J.E. Jarrett), xpan1@uis.edu (X. Pan).
0167-9473/$ - see front matter 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.csda.2006.01.020
J.E. Jarrett, X. Pan / Computational Statistics & Data Analysis 51 (2007) 38623870 3863
autocorrelated data. This is important because the i.i.d. assumptions of standard Shewhart control charts do not hold
for large amounts of historical data.
Kalgonda and Kulkarni (2003, 2004) considered using the Hotelling T
2
statistic to monitor several quality char-
acteristics. Due to difculties of this statistic in determining the factors responsible for out of control signals when
considering changes in the process mean vector and/or process covariance matrix, they proposed a new diagnostic,
the D-technique. Their analysis including a dummy variable multiple linear regression provided some evidence of
usefulness of constructing multivariate control charts. Their second study (Kalgonda and Kulkarni, 2004) extended the
usefulness of multivariate control charts in detecting an out of control status and identifying variable(s) responsible for
the out of control situation.
Another study, Jaerkaporn et al. (2003) corroborated the usefulness of using average run length methods to the
standard Shewhart control charts for individuals. Their method showed promise that average run length control charts
are superior for detecting shifts in the mean than the traditional charts. Sliwa and Schmid (2005) extended the use
of average run length to multivariate time series. Their study focused on the autocovariance and the cross-covariance
structure of nancial assets. Hence, monitoring the process control for new multivariate time series resulted from their
approach. Finally, Cheng and Thaga (2005) produced a CUSUM control chart capable of detecting changes in the
mean and standard deviation of autocorrelated data. Based on the average run length, they claim that their new control
chart (MCAP) is useful for monitoring modern production processes where one produces high quality goods with a
tiny fraction of nonconforming output units.
We propose to continue the study of multivariate autocorrelated processes by proposing a vector autoregressive (VAR)
control chart based on ARL for precision manufactured products that are now produced en masse in high technology
oriented industries. One example of such production is optical communication products manufacturing where the
output is a multivariate autocorrelated time series. One bases the production of ber optic on SiO
2
rod made from
condensation of silicon and oxygen gases. The preparation of the SiO
2
rod needs to control temperature, pressure and
the concentrations of different components. We nd similar processes in semiconductors where they prepare materials.
In these processes, correlated variables are plentiful and the addition of multivariate analysis is necessary to monitor the
process. In addition, these variables are cross-correlated with time leads and lags. Univariate quality control charts are
effective for monitoring individually correlated variables but are not effective for services, which are cross-correlated.
We propose to introduce (VAR) control charts designed to monitor multivariate processes of the type described.
Shewhart control charts do not emphasize parameter estimation. This lack of concern emanates from using the rough
criterion of average run length (ARL) to determine effectiveness. For autocorrelated processes, estimation is the key
procedure for control chart construction. Some of the proposed control charts for autocorrelation are also not applicable
if we ignore estimation. Hence, we will consider estimation issues associated with ARIMA modeling.
We arrange the remaining parts of this paper as follows: In Section 2, we describe the principles and design of VAR
control charts. In Section 3, we examine the effects of parameter shifts. In Section 4, we show an example VAR chart
construction for a multiple variable AR (3) process. In turn, we examine the ARL. Last, in Section 5, we summarize.
2. VAR control chart
We propose using aVARcontrol chart (another title is MultivariateAutoregression (MAR) control chart). The process
is required either stationary or we can achieve stationarity with the appropriate ltering process, i.e. differencing. For
a multivariate autoregressive process of n variables, denote y
t
= (y
1t
, y
2t
, . . . , y
nt
)

as a (n 1) vector. These cross-

correlated variables contain autocorrelation of order p. The process follows a VAR model
y
t
= c +

1
1
L + 1
2
L
2
+ + 1
p
L
p

y
t
+ e
t
= c + 1(L) y
t
+ e
t
, (1)
where L is the backshift operator, c = (c
1
, c
2
, . . . , c
n
)

is the constant vector, and e

t
= (e
1t
, e
2t
, . . . , e
nt
)

is the vector
for the error term. Each 1
j
is a (n n) coefcient matrix for lag j. e
t
contain no autocorrelation but does contain
cross-sectional correlation. That is, E

e
t
e

= O is invariant about time but may not be a diagonal (n n) matrix.

Some elements of the matrix 1
j
are zero when variables are uncorrelated.
In this model, the AR coefcient matrix, 1 =

1
1
1
2
1
p

, and the constant term c represent systematic patterns.

The error termvector e
t
represents nonsystematic disturbances. We monitor the process to lter the systematic element.
In turn, we monitor the special-cause. To lter out and estimate the systematic element for a process that is in-control
3864 J.E. Jarrett, X. Pan / Computational Statistics & Data Analysis 51 (2007) 38623870
requires multivariate regression, which, in turn, requires huge sample sizes for asymptotic estimation. To construct
control charts by these methods, we implement a two-stage routine. For in-control cases, we estimate the systematic
elements. We denote H

=( c 1) being an n(1+np) matrix, and x

t
=

1, y

t 1
, y

t 2
, . . . , y

t p

being a 1(1+np)
vector. Then Eq. (1) becomes
y
t
= H

x
t
+ e
t
. (1

)
Due to the usual assumptions about e
t
, we estimate the coefcient matrix n

by ordinary least squares (OLS). For

Gaussian AR(p) process, e
t
i.i.d N

0, O

, these OLS estimates are

t =1
y
t
x

t =1
x
t
x

1
, (2)
where T is the number of observations in Stage I. The estimated systematic model is

y
t
=

H

x
t
. (3)
In addition, we estimate error term as

e
t
= y
t


y
t
. We then develop the special-cause control chart based on e
t
.
Maximum likelihood methods estimate the covariance of e
t
as

O = (1/T )
T

t =1

e
t

e

t
p
O. (4)
If the process is in-control, the model in (1) is appropriate and the error termis uncorrelated and normal, the estimated
error e
t
should be also approximately normal and invariant with time. Thus, it is asymptotically N

0, O

. We now
apply the multivariate control chart e
t
. In Stage I the Hotelling T
2
statistic is, approximately ,
2
(n) when T n is large.
T
2
=

e
t

O
1

e
t

,
2
(n), (5)
where

e is mean of Stage I,

e =

T
t =1

e
t
/T , which should be close to zero. Since the estimates of the AR(p) model are
asymptotically available, we use chi-square instead of Beta distribution.
For Stage II observations, Y

i
, (i >T ), we denote the rst Stage II observation as the (T + 1)st observation. We
substitute the estimated error by the one-step-ahead forecast error, while we compare the rst Stage II observation with
the one-step-ahead forecast from Stage I. That is, from the known T observations in Stage I, we predict the rst Stage
II status as

y
T +1
=

H

x
T +1
. (6)
Moreover, the forecasting error is the difference between the real observation and the forecast,

e
T +1
=

y
T +1
y
T +1
. (7)
There are several possible schemes for Stage II chart. If we treat

O as Wishart distribution, the chart statistic is
T
2
T +1
=

e
T +1

O
1

e
T +1

n(T + 1)(T 1)
T (T n)
F
:,n,T n
. (8)
However, in an autoregression environment, we only have asymptotic properties,

O O.
Therefore, the more appropriate way is to use the approximation:
T
2
T +1
=

e
T +1

O
1

e
T +1

,
2
(n). (9)
J.E. Jarrett, X. Pan / Computational Statistics & Data Analysis 51 (2007) 38623870 3865
For the following observations, the forecasting errors are acquired recursively one-step-ahead each. Generally, for
the forecasting error of the ith observation,

e
i
=

H

x
i
y
i
, the chart statistic is
T
2
i
=

e
i

O
1

e
i

,
2
(n). (10)
An F distribution is asymptotically close to a chi-square distribution (divided by T ). The asymptotic distributions of
the estimates provide, under the same condence level, more narrow condent intervals for true parameters than the
exact distribution for limited sample size T. Therefore, control charts using chi-square distributions are more sensitive
than those using F distributions.
Selecting statistic: One statistic is given in (10). By selecting the expression of the mean itemof T
2
i
we can have more
chart schemes. One option is directly choosing

e as its asymptotic value, zero. Then the item

e in (10) will disappear:

T
2
i
=

e

O
1

e
i
,
2
(n). (11)
The corresponding Stage I statistic is
T
2
=

e

t

O
1

e
t
,
2
(n). (5

)
Another option is using recursive mean. This is corresponding to the chart of moving centerline in univariate case
(Montgomery and Mastrangelo, 1991). The mean for the (i + 1)th observations statistic can be dened as

e
i
=

e
i
+ (i 1)

e
i1

/i so that the statistic is

T
2
i
=

e
i

e
i

O
1

e
i

e
i

,
2
(n). (12)
Although the recursive mean varies every time when a new observation is added, the control limit of the chart does
not change. This is much better than the moving center line in the univariate case. Note, all the above schemes have
the same chart that we determine the control limit (UCL) by ,
2
(n) say, ,
2
0.01
(n). Only the statistic values are different
for different schemes. Hence, the VAR chart is actually simple in its demonstration, and model estimation is relatively
simple. Among (10)(12), the statistic in (11) is the most convenient. Because the asymptotic mean of

e
t
is zero thus
(11) is theoretically justied for large sample size.
Determining the AR order: In practice, determining the autocorrelation order p may be the issue before chart setting.
To nd a parsimonious AR order, the null hypothesis is that the AR order is p
0
as opposed to being p
1
>p
0
. The test
statistic is a likelihood ratio:
2 (L
1
L
0
) = T

log

O
0

log

O
1

,
2

n
2
(p
1
p
0
)

, (13)
where T is the sample size in Stage I. O
0
is the estimated covariance matrix of the error term given in (10) under the
null hypothesis being true. O
1
is also from (10) but under the alternative hypothesis. The chi-square distribution has a
degree of freedom n
2
(p
1
p
0
) because each variable has (p
1
p
0
) less lags under the null hypothesis and each of
the n equations include n variables and totally n
2
restrictions.
3. The effect of shift
There are three classes of parameters in VAR models: the process mean j, the shift on covariance matrix of error
term, O, and the shift autocorrelation structure of the model, 1. The most important shift of interest is the shift on the
mean j of the process.
For a mean shift, the relationship between the mean of a stationary time series vector, j is the expectation of Eq. (1).
The following theorem gives the effects of mean shift on the VAR chart statistic. There are possibly two ways for a shift
to build on the ARMA process. One is that the shift is added on the constant item, meaning there is really a physical
3866 J.E. Jarrett, X. Pan / Computational Statistics & Data Analysis 51 (2007) 38623870
structure that the historical statuses are added on the constant. Another way is that the shift is added on the mean of the
process, which is more an observed result. We here only consider the rst situation.
Theorem 1. For a process model in (1) and VAR chart in (10), a shift

o in the process constant adds a normal
distribution with mean of

n

O
1

n
k
and variance of 4

k

O
1

n
k
on the in-control T
2
. The effects depend on the shift
magnitude and the estimates of VAR parameter matrices, n = (I
n


1(L))

o.
Proof. In the following proof, we use the estimated coefcients rather than the known coefcients. Denote the in-
control residual here as

e
t
, and the out-of-control (due to shift) residuals as

c
t
. In addition, we use z
t
to denote the
process after shift. These changes in denotation are not necessary but just for an alternative expression. The results
for the shift effects are the same except that one is univariate ARMA and the other is multivariate AR, because the
asymptotic estimates should be just the known parameters.
The forecasting error without shift occurrence is

e
t
= y
t


y
t
. After the shift, the process status is z
t
rather than y
t
.
A sudden shift at time just before t
0
on mean for some of the variables implies
z
t
= y
t
+

o for t t
0
.
Since the shift is unknown before detection, we still base the one-step-ahead forecasting on the estimated coefcient
matrix

H

c

1

. Thus, for a VAR(p) process,

z
t
=

c +

1(L) z
t
when t t
0
+ p.
Then the forecasting error after the shift is

t
t
= z
t


z
t
= z
t


c

1(L) z
t
= y
t
+

o

c

1(L) y
t


1(L)

o
= y
t


1(L)

o

y
t
+

o =

e +

I
n


1(L)

o
=

e
t
+

n.
Here the

e
t
was the forecasting error at time t, had the shift not occurred. With the shifted


t
=

e
t


n, the Hotelling
T
2
statistic will be a noncentral chi-square distribution:
shift
T
2
t
=

t

O
1

t
=

e
t


n

O
1

e
t


n

=

e

t

O
1

e
t
2

O
1

e
t
+

n

O
1

n
=
noshift
T
2
t
2

O
1

e
t
+

n

O
1

n.
The rst term is the in-control Hotelling statistic
noshift
T
2
t
,
2
(n) with mean of n and variance of 2n. Since
asymptotically

e
t
N

0, O

, the second term 2

O
1

e
t
is a linear combination of normal distributions so that it
is normal with mean of 0 and variance of 4

O
1

O

O
1

n = 4

O
1

n. Its effect, we view that the rst term increases

variation but without shifting the mean. The third term is a positive denite scalar. The last two terms are normal
distribution with mean of

n

O
1

n and variance of 4

O
1

n. Therefore, we measure the total effect of the process shift

n on T
2
by

n

O
1

n.
When t
0
t <t
0
+ p, let k = t t
0
. We have z
t
= y
t
+

o for t
0
t <k and k lags shift while p k lags do not shift.
The one-step-ahead prediction is

z
t
=

c +

1
1
z
t 1
+

1
2
z
t 2
+ +

1
k
z
t k
+

1
k+1
y
t k1
+ +

1
p
y
t p
=

c +

1
1
L +

1
2
L
2
+ +

1
k
L
k

z
t
+

1
k+1
y
t k1
+ +

1
p
y
t p
.
J.E. Jarrett, X. Pan / Computational Statistics & Data Analysis 51 (2007) 38623870 3867
Thus

t
t
= z
t


z
t
= z
t

c +

1
1
L +

1
2
L
2
+ +

1
k
L
k

z
t
+

1
k+1
y
t k1
+ +

1
p
y
t p

= y
t
+

o

c +

1
1
L +

1
2
L
2
+ +

1
k
L
k

y
t
+

1
k+1
y
t k1
+ +

1
p
y
t p

1
1
L +

1
2
L
2
+ +

1
k
L
k

o
=

e
t
+

n
k
,
where

n
k
=

I
n


1
1


1
2


1
k

o and k = 0, 1, 2, . . . , (t t
0
).
Therefore the effect on VAR chart statistic is, for t
0
t t
0
+ p,
shift
T
2
t
=
noshift
T
2
t
2

O
1

e
t
+

n

O
1

n
k
,
where k =t t
0
. This is still a positive shift on the mean of the in-control T
2
and on the variance of the in-control T
2
.
From the denition of

n and

n
k
, obviously

n =

n
p
for t t
0
+ p, because 1
k
= 0 for k >p, and the estimates of 1
k
should have the same property asymptotically. The effects during t
0
t t
0
+p can be either larger or smaller than the
effects during t t
0
+ p, depending on the specic VAR structure

1(L) and

O.
Other systematic shifts, such as the shift on the covariance matrix of error term, O, or the shift on the autocorrelation
structure of the model, 1, will also violate the distributions of the chart statistic, hence out of control signals will be
observed after the shifts. The larger the shift scale, the more serious the violation of the chi-square distribution. The
out-of-control ARL, which determines when the VAR chart detects the shift, relates to the control limit and the scale
of the shift.
4. A simulated example of a VAR chart
Following is an illustration showing the procedures and properties of a VAR chart. We specify three as both the order
of autocorrelation and the number of variables. We design the covariance matrix of the error term vector as
O =

1 0.5 1.5
0.5 4.25 1.75
1.5 1.75 2.99

.
In this way, we keep the positive denite. The constant term is set to be (3

= (1.1 2.1 1.1), and the three coefcient

matrices are as follows:
1
1
=

0.5 0.2 0.08

0.1 0.7 0.2
0.25 0.22 0.54

, 1
2
=

0.25 0.23 0.1

0.34 0.32 0.12
0.43 0.18 0.15

, 1
3
=

0.21 0.1 0.13

0.09 0.32 0.21
0.15 0.05 0.48

.
By this design, all of the characteristic values of

I
n
z
p
1
1
z
p1
1
2
z
p2
1
p

= 0, lie in the unit circle,

|z
i
| <1, i =1, 2, . . . , p. This setting guarantees a stationary model (p =3, n=3). In addition, the variable means are
j

= c

I
n
1
1
L 1
2
L
2
1
3
L
3

= (0.4982 11.6499 15.082). With the above parameters, we sampled and

experimented using Matlab. Fig. 1(a) and (b) are the results for the two statistics in (5) and (5

) at Stage I, where the

UCL was preset at : = 0.005. From the gures, all the 100 observations fell within the UCL. Actually, if we repeat
the same simulation, sometimes we found one or two observations fallout of UCL. This may imply the in-control ARL
may be suitable. In fact, in Fig. 1(c) and (d) if we try 500 observations, we observed from one to ve out UCL samples.
Fig. 1 also showed that the statistics in (11) and (11

) at Stage II almost have no difference.

To conrmthe autocorrelation order p, we run the likelihood ratio test in (13) for the hypotheses (p
0
= 2 vs. p
1
= 3),
(p
0
= 3 vs. p
1
= 4), (p
0
= 3 vs. p
1
= 6), and (p
0
= 4 vs. p
1
= 6). At : = 0.05 the null hypothesis was rejected for
rst test and cannot be rejected for the other three tests. This showed that the recommended method for selecting order p
3868 J.E. Jarrett, X. Pan / Computational Statistics & Data Analysis 51 (2007) 38623870
0 50 100 150 200 250 300 350 400 450 500
0
2
4
6
8
10
12
14
16
18
STAGE I VAR CHART WITH
MEAN ERRORS, alpha=0.005
Observations
H
o
t
e
l
l
i
n
g

T
-
s
q
u
a
r
e

S
t
a
t
i
s
t
i
c
0 50 100 150 200 250 300 350 400 450 500
0
2
4
6
8
10
12
14
16
18
STAGE I VAR CHART WITH
ZERO MEAN, alpha=0.0027
Observations
H
o
t
e
l
l
i
n
g

T
-
s
q
u
a
r
e

S
t
a
t
i
s
t
i
c
0 10 20 30 40 50 60 70 80 90 100
0
2
4
6
8
10
12
14
STAGE I VAR CHART WITH
ZERO MEAN, alpha=0.005
Observations
H
o
t
e
l
l
i
n
g

T
-
s
q
u
a
r
e

S
t
a
t
i
s
t
i
c
0 10 20 30 40 50 60 70 80 90 100
0
2
4
6
8
10
12
14
STAGE I VAR CHART WITH MEAN
ERRORS, alpha=0.005
Observations
H
o
t
e
l
l
i
n
g

T
-
s
q
u
a
r
e

S
t
a
t
i
s
t
i
c
(a)
(b)
(c)
(d)
Fig. 1. (a) Stage | VAR chart with mean errors, alpha = 0.005. (b) Stage | VAR chart with zero mean, alpha = 0.005. (c) Stage | VAR chart with
mean errors, alpha = 0.005. (d) Stage | VAR chart with zero mean, alpha = 0.0027.
is valid for this model. It should be noted that not rejecting hypothesis (p
0
= 4 vs. p
1
= 6) does not mean the selection
is nished. Continuing searches occur until we reject the null hypothesis for (p
0
= 2 vs. p
1
= 3).
We examine the in-control ARL via experimental simulation. Observe the histogram of the run length distribution
based on certain estimated in-control process parameters in Fig. 2(a). The ARL is 215 for :=0.0027, and the respective
standard deviation is 227. Hence, the UCL should be set higher than anticipated. Fig. 2(b) shows the run length
distribution at higher UCL, : = 0.0020. This ARL is 282.6 and the standard deviation is 305.3.
When a shift occurs on the variable means of the process, say, the constant term of the rst and the third variable,
c
1
and c
3
, are doubled, then c

= (2.2 2.1 2.2), and the mean vector becomes 1

= [0.6198 18.1950 27.5453]. By

comparison, this is a very large shift in mean and covariance matrix (about a one error-term sigma shift on Variable
1, half error-term-sigma shift on Variable 2, and four sigma shift in Variable 3). The result of the shift is the ARL =
1.24 and its standard deviation is 0.56. (: =0.002). The VAR chart is thus very sensitive to shifts as shown in Fig. 3 at
observation 80.
J.E. Jarrett, X. Pan / Computational Statistics & Data Analysis 51 (2007) 38623870 3869
100 300 500 700 900 1100 1300 1500
0
10
20
30
40
50
60
Histogram of Run Length Distribution
Run Length
F
r
e
q
u
e
n
c
y
50 150 250 350 450 550 650 750 850 950 10501150125013501450
0
50
100
150
200
250
300
350
400
Histogram of Run Length Distribution
Run Length
F
r
e
q
u
e
n
c
y
(a)
(b)
Fig. 2. (a) Histogram of run length distribution. (b) Histogram of run length distribution at higher UCL.
0 10 20 30 40 50 60 70 80 90 100
0
10
20
30
40
50
60
Run Chart for Shift
Observations
T
-
s
q
u
a
r
e

S
t
a
t
i
s
t
i
c
Fig. 3. Run chart for shift.
If a small shift occurs, for example, the constant term of the third variable increases by 50%, (corresponding to about
0.4 error-term-sigma shift on Variable 2 and 0.9 error-term-sigma shift on Variable 3), the ARL is 104 and standard
deviation of ARL is 117 (a = 0.0020). Similarly, if the constant of the third variable decreases 50%, the effect on
ARL is 113. Similar to the increase case, although the shift on the means are different (half error-term-sigma shift on
Variable 1 and one error-term-sigma shift on Variable 3).
3870 J.E. Jarrett, X. Pan / Computational Statistics & Data Analysis 51 (2007) 38623870
Since the simulations were based on certain estimates of the parameters instead of the true underlying parameters,
the ARL values are different from the results using true parameters. For example, when true parameters were used, the
mean shift effect in above give ARL =84.5 with 8.67 standard error. ARL based on estimates may be larger or smaller
than the ARL from true parameters, depending on whether overvaluing or undervaluing the estimates. As the system is
dynamic, a given set of data can produce overestimation on one parameter and underestimation on the other. Since the
true process is unknown, there is actually no way to assess if one overvalues or undervalues the ARL. The only prudent
way is to collect as many data as possible in Phase I to acquire asymptotic estimates.
5. Summary
Var control charts are a combination of univariate residual chart for autocorrelation and multivariate chart for
independent processes. Based on our development and carefully explained methodology, one can model a multivariate
process by AR equations describing its physical properties and estimate the AR(p) model by OLS procedures, and
applies VAR control chart to monitoring autocorrelated multivariate processes. Hence, the VAR control chart is both
useful as long as sample size are sufcient for OLS estimation.
References
Alwan, B.M., Roberts, H.V., 1988. Time-series modeling for detecting level shifts of autocorrelated processes. J. Bus. Econ. Stat. 6 (1), 8796.
Cheng, S.W., Thaga, K., 2005. Max-Cusum chart for autocorrelated processes. Stat. Sin. 15, 527546.
English, J.R., Sastri, T., 1990. Enhanced quality control in continuous ow processes. Comput. Ind. Eng. 19, 258262.
Hawkins, D.M., 1991. Multivariate quality control based on regression-adjusted variables. Technometrics 33 (1), 6175.
Hawkins, D.M., 1993. Regression adjustment for variables in multivariate quality control. J. Qual. Technol. 25(3), 170182.
Jaerkaporn, D., et al., 2003. Process monitoring for correlated gamma-distributed data using generalized-linear-model-based control charts. Qual.
Rel. Eng. Int. 19 (6), 477491.
Kalgonda, A.A., Kulkarni, S.R., 2003. Diagnosis of multivariate control chart signal based on dummy variable regression technique. Commun. Stat.:
Theory and Methods 32 (8), 16651684.
Kalgonda, A.A., Kulkarni, S.R., 2004. Multivariate quality control chart for autocorrelated process. J. Appl. Stat. 31 (3), 317327.
Loredo, E.N., Jaerkpaporn, D., Borror, C.M., 2002. Model-based control chart for autoregressive and correlated data. Qual. Rel. Eng. Int. 18 (6),
489496.
Lu, C.W., Reynolds, M.R., 1999. Control charts for monitoring the mean and variance of autocorrelated processes. J. Qual. Technol. 31, 259274.
Montgomery, D.C., Mastrangelo, C.M., 1991. Some statistical process control methods for autocorrelated data. J. Qual. Technol. 23 (3), 179193.
Sliwa, P.L., Schmid, W., 2005. Monitoring the cross-covariances of a multivariate time series. Metrika 61 (1), 89115.
Snoussi, A., Ghourabi, M.E., Limam, M., 2005. On SPC for short run autocorrelated data. Commun. Stat.: Simulation and Computation 34 (1),
219234.
Testik, M.C., 2005. Model inadequacy and residuals control charts for autocorrelated processes. Qual. Rel. Eng. Int. 21 (2), 115130.
Wardell, D.G., Moskowitz, H., Plante, R.D., 1994. Run-length distribution of special-cause control charts for correlated processes. Technometrics
36, 10841105.