Monitors a single curve that is only partially observed on the domain. The unobserved portion is completed using one of several methods before projecting onto the Phase I FPCA basis.
Usage
spm.monitor.partial(
chart,
partial.data,
n.observed,
completion = c("conditional", "projection", "zero_pad"),
ncomp = NULL,
alpha = 0.05
)Arguments
- chart
An object of class
spm.chartfromspm.phase1.- partial.data
Numeric vector of observed values for one curve.
- n.observed
Integer; how many of the grid points have been observed (from the start of the domain).
- completion
Completion method:
"conditional"(conditional expectation / BLUP),"projection"(partial projection), or"zero_pad".- ncomp
Number of components (default: same as chart).
- alpha
Significance level (default 0.05).
Value
A list with components:
- scores
Estimated FPC scores
- t2
T-squared statistic
- t2.alarm
Logical: TRUE if T-squared exceeds UCL
- domain.fraction
Fraction of domain observed
- completed.curve
The completed curve (full domain), or NULL
See also
spm.monitor for fully-observed curves,
spm.monitor.partial.batch for multiple partial curves
Examples
# \donttest{
set.seed(1)
n <- 50; m <- 30
argvals <- seq(0, 1, length.out = m)
X <- matrix(rnorm(n * m), n, m)
fd <- fdata(X, argvals = argvals)
chart <- spm.phase1(fd, ncomp = 3)
# Monitor a partially observed curve (first 20 of 30 grid points)
partial <- rnorm(30)
result <- spm.monitor.partial(chart, partial, n.observed = 20)
result$t2
#> [1] 1.60075
result$domain.fraction
#> [1] 0.6551724
# }