The following is an illustration for EWMA charts, assuming that all observations are normally distributed.
Based on \(n\) past in-control observations \(X_{-n},\dots,X_{-1}\), the in-control mean and standard deviation can be estimated by the sample mean, \(\hat \mu\), and sample standard deviation, \(\hat \sigma\). For new observations \(X_1,X_2,\dots\), an EWMA chart based on these estimated parameters is defined by \[ M_0=0, \quad M_t=\lambda \frac{X_t-\hat \mu}{\hat \sigma}+(1-\lambda) M_{t-1} \]
The following generates a data set of past observations (replace this with your observed past data).
X <- rnorm(250)
Next, we initialise the chart and compute the estimates needed for running the chart - in this case \(\hat \mu\) and \(\hat \sigma\).
library(spcadjust)
chart <- new("SPCEWMA",model=SPCModelNormal(Delta=0),lambda=0.1);
xihat <- xiofdata(chart,X)
str(xihat)
## List of 3
## $ mu: num 0.0251
## $ sd: num 1.05
## $ m : int 250
We now compute a threshold that with roughly 90% probability results in an average run length of at least 100 in control. This is based on parametric resampling assuming normality of the observations.
cal <- SPCproperty(data=X,nrep=50,
property="calARL",chart=chart,params=list(target=100),quiet=TRUE)
cal
## 90 % CI: A threshold of +/- 0.5342 gives an in-control ARL of at
## least 100.
## Unadjusted result: 0.4928
## Based on 50 bootstrap repetitions.
You should increase the number of bootstrap replications (the argument nrep) for real applications.
Next, we run the chart with new observations that are in-control.
newX <- rnorm(100)
S <- runchart(chart, newdata=newX,xi=xihat)
Then we plot the data and the chart.
par(mfrow=c(1,2),mar=c(4,5,0.1,0.1))
plot(newX,xlab="t")
plot(S,ylab=expression(S[t]),xlab="t",type="b",ylim=range(-cal@res,S,cal@res+0.3,cal@raw))
lines(c(0,100),rep(cal@res,2),col="red")
lines(c(0,100),rep(cal@raw,2),col="blue")
abline(0,0,lty=3)
lines(c(0,100),rep(-cal@res,2),col="red")
lines(c(0,100),rep(-cal@raw,2),col="blue")
legend("topleft",c("Adjusted Threshold","Unadjusted Threshold"),col=c("red","blue"),lty=1)
In the next example, the chart is run with data that are out-of-control from time 51 and onwards.
newX <- rnorm(100,mean=c(rep(0,50),rep(-1,50)))
S <- runchart(chart, newdata=newX,xi=xihat)
