The historical percentage yields from a batch reactor for 300 sequential batches.
Data source
Then conditions were changed to try improve the yield by running 10 new batches, and these yields were recorded: \([83.5, 78.9, 82.7, 93.2, 86.3, 74.7, 81.6, 92.4, 83.6, 72.4]\). These values are not included in the downloadable dataset.
The 300 yield values may be used to create a historical reference set of typical variation, allowing one to compare whether the 10 new batches actually led to a significant improvement.
Simulated data in R:
N <- 300
phi <- -0.3
spread <- 6.7
location <- 79.9
y <- numeric(N) # create a vector of zeros
for (k in 2:N)
{
y[k] <- phi*y[k-1] + rnorm(1, mean=0, sd=spread)
}
y <- y + location