Hi!
Thank you for your great efforts here!

I was wondering whether there is a function contained in this package to perform Levey-Jennings-Chart?

Best,
Payam

Hi,

I recently updated some data that I have been running through the QC_Violations function. For some reason a particular set of values seems to trigger the following error message:

> Error in `$<-.data.frame`(`*tmp*`, "Violation_Same_Side", value = c(TRUE,  : 
>   replacement has 10 rows, data has 20

I've been unable to ascertain exactly what is causing the error, but am able to consistently reproduce it.

# Values which originally triggered the error msg
x1 <- c(
         0.728358208955224, 0.69198606271777, 0.696148359486448, 0.725755166931638, 
         0.729329173166927, 0.718320610687023, 0.70610119047619, 0.724165341812401, 
         0.742485783915516, 0.730029585798817, 0.798306389530408, 0.818871103622578, 
         0.769933554817276, 0.777939042089985, 0.814540059347181, 0.835034013605442, 
         0.789566755083996, 0.802180685358255, 0.802058590657165, 0.812206572769953
         )

QC_Violations(data=x1, method = "XmR")


# Changing the last value to 0.999 stops the error from flagging
x2 <- c(
  0.728358208955224, 0.69198606271777, 0.696148359486448, 0.725755166931638, 
  0.729329173166927, 0.718320610687023, 0.70610119047619, 0.724165341812401, 
  0.742485783915516, 0.730029585798817, 0.798306389530408, 0.818871103622578, 
  0.769933554817276, 0.777939042089985, 0.814540059347181, 0.835034013605442, 
  0.789566755083996, 0.802180685358255, 0.802058590657165, 0.999
)

QC_Violations(data=x2, method = "XmR")


# random values seem to work
QC_Violations(data=rnorm(20), method = "XmR")

# sequential values don't
QC_Violations(data=as.numeric(1:20), method = "XmR")

# sometimes adding random values to sequential values works, sometimes it doesn't
QC_Violations(data=as.numeric(1:20) + rnorm(20), method = "XmR")

There is some discussion of this error message here, however I haven't been able to figure out how that applies to this scenario.

I'm running ggQC version 0.0.3 on R 4.3.2, using RStudio 2023.06.0 Build 421. on a Windows 10 machine.

When using ggQC I get multiple upper and lower control lines,

Here is the code, which I use

# Socket Thickness Data
file.path = '~/Documents/R-Programs/WheelerDonald/'
file.name = 'data_Wheeler_SocketThickness.csv'
DF = read.table(paste0(file.path, file.name), sep=",", header = TRUE)

##
# Combine time and cycles and generate a new column:
DF$subGrp = paste0(DF$time, DF$cycles)
ggplot(DF, aes(x=subGrp, y=y, color=color)) + stat_QC(method="xBar.rBar")

I also tried to convert "subGrp" into a factor, but the result is the same. I append the dataset.

data_Wheeler_SocketThickness.csv

For example, when setting LSL=NULL as below, the calculated Cpk is 7.39. However, the correct Cpk should be 2.46. Could the author help to review the code?

CapabilityAnaylsis <-
ggplot(Process_Data, aes(x = Value)) + #init ggplot
geom_histogram(binwidth = .75, color="purple") + #make the histogram
stat_QC_Capability(
LSL=NULL, USL=37, #Specify LSL and USL
show.cap.summary = c("Cp", "Cpk", "Pp, "Ppk"), #selected summary
digits = 2, #report two digits
method="XmR") + #Use the XmR method
scale_x_continuous(expand = expand_scale(mult = c(0.15,.65))) #pad the X-axis
CapabilityAnaylsis

It seems XBar functions and graphs ignore subgroup sizes (unweighted mean, std dev etc are used - e.g. in XCentral_LimitFUN in the source). It would be great if this could be incorporated.

Hi, Sr. thank you for this package.

Apparently, the code is running, but I'm having some problems with the order of the frequency in the chart.

e.g( holes has the highest value but it's the scratch that appears first on the chart)

library(tidyverse)
library(ggQC)

df <- tibble(type_nc = c("scratch","spots","folds","holes","rips"),
                       qty = c(201,78,47,31,15),
                       cost = c(4287,2423,1119,8947,1864))
df

df %>% 
  ggplot(aes(x = type_nc,
             y= cost)) + 
  stat_pareto(point.color = "navy",
              point.size = 2,
              line.color = "navy",
              bars.fill = c("blue","lightblue"))

Thank you

How can add a grid only for the secondary axis (cummulative %)?

Warning:
The fun.y argument of stat_summary() is deprecated as of ggplot2 3.3.0.
ℹ Please use the fun argument instead.
ℹ The deprecated feature was likely used in the ggQC package.
Please report the issue to the authors.

Backtrace:
▆

1. └─ggQC::stat_QC(method = "xBar.rBar", auto.label = T, label.digits = 2)
2. └─ggplot2::stat_summary(...)

3. └─ggplot2:::deprecate_warn0("3.3.0", "stat_summary(fun.y)", "stat_summary(fun)")
   

While the bars show up in the correct order (largest to smallest), their labels are incorrect any time x is not in alphabetical order. Could this be fixed?

# first set alphabetized, second not, but with same value order
sorted     <- data.frame(label = c("A",   "B",   "C",     "D"),    value = c(20, 15, 10, 5))    
not_sorted <- data.frame(label = c("One", "Two", "Three", "Four"), value = c(20, 15, 10, 5))    

# this one labels bars correctly
ggplot(sorted, aes(x = label, y = value)) +

    ggQC::stat_pareto(point.color = "red",
                point.size = 2,
                line.color = "black",
                bars.fill = c("coral", "yellow"))
    
# this one does not label bars correctly
ggplot(not_sorted, aes(x = label, y = value)) +

    ggQC::stat_pareto(point.color = "red",
                point.size = 2,
                line.color = "black",
                bars.fill = c("coral", "yellow"))

The image below shows black points inside green points on an X-bar chart built by following the example in this vignette (plot title removed). The green larger points were set by my code (below). However the default black points persist. I have changed the order of layers setting individual measurement points first (faded gray) followed by average measurements green points with zero transparency and the faded points seem to have been masked by the green points.
The black points together with the lines however, are not masked by the green points.

ggQC package version: 0.0.31

Thank you!

Xbar = lapply(unique(VectorData), function(i) {
       ggplot(dtt[Item %in% i], aes(x = Sample, y = .data[['Area/\u03BCL']], group = 1L)) +
       theme_bw() + geom_point(alpha = 0.2, size = 2) +
       stat_summary(fun = 'mean', geom = 'point', size = 4, color = '#17b903') + 
       stat_summary(fun = 'mean', geom = 'line') + 
       theme(axis.text.x = element_text(size = 6, angle = 90, vjust = 0.5, hjust=1)) +
       ggtitle(sprintf('%s  | Item : %s', equip, i)) + 
       scale_x_discrete(expand = expansion(add = c(1L, 2L))) +
       stat_QC() + stat_QC_labels(text.size = 2)
    })

Using stat_summary() with argument fun does not prevent the

Warning message:
The fun.y argument of stat_summary() is deprecated as of ggplot2 3.3.0.
ℹ Please use the fun argument instead.
ℹ The deprecated feature was likely used in the ggQC package.
Please report the issue to the authors.
This warning is displayed once every 8 hours.
Call lifecycle::last_lifecycle_warnings() to see where this warning was generated:

Backtrace:

1. base::lapply(...)
2. global FUN(X[[i]], ...)
3. ggQC::stat_QC()
4. ggplot2::stat_summary(...)
5. ggplot2:::deprecate_warn0("3.3.0", "stat_summary(fun.y)", "stat_summary(fun)")

I am rising this issue for the reason that package `ggQC` is mentioned in the warning. 

`ggQC` package version: 0.0.31

Thank you!

When trying to plot new data on the set control limit lines, if you use anything but the XmR option, the control limits only show for the "Setup"/"pre_change"

#Data Generation
dat_i <- data.frame(mnth = 1:24,
r_cnt = rbinom(24,100,prob = p_red),
DataType = "pre_change"
)
dat_i$r_p = dat_i$r_cnt/100

dat_n <-
data.frame(mnth = 25:36,
r_cnt = rbinom(24,100,prob = .15),
DataType = "post_change"
)
dat_n$r_p = dat_n$r_cnt/100
dat_all <- rbind(dat_i,dat_n)

#Plotting the Data
P_Plot <-
ggplot(dat_all, aes(x = mnth, y = r_p, n = 100)) +
geom_point() + geom_line() +
stat_QC(data = dat_all[dat_all$DataType=="pre_change",],
method = "p",
auto.label = T,
label.digits = 2,
show.1n2.sigma = T
) +
scale_x_continuous(expand = expansion(mult = .15))

P_Plot

Hello,
Thank you in advance for your help! I am utilizing your package and I am having an issue with the point color for qc violations. I was hoping you would have an idea of what I am doing wrong.

m_base_count <- ggplot(diehard, aes(x = Date Opened, y = n)) + stat_qc_violations(method = "XmR", violation_point.color = "Red") + ggtitle("All Violations") m_base_count

Please let me know if the picture does not work. I am very to Github.

The code for this Issue is posted in Issue #51:

Using argument size in stat_summary(geom = 'line', size = 1) throws the following warning:

<warning/lifecycle_warning_deprecated>
Warning:
Using `size` aesthetic for lines was deprecated in ggplot2 3.4.0.
ℹ Please use `linewidth` instead.
---
Backtrace:
 1. ggplot2::stat_summary(...)
 2. ggplot2::layer(...)
 3. ggplot2:::deprecate_soft0(...)

However, changing the argument to stat_summary(geom = 'line', lineWidth = 1) throws the following

Warning message:

In stat_summary(fun = "mean", geom = "line", lineWidth = 0.5, color = "gray90") :
  Ignoring unknown parameters: `lineWidth`

ggQC package version: 0.0.31

Thank you!

When running the following code, I got warning message "fun.y is deprecated"
After dumping the function code, I think is cause by stat_QC.
How can I fix this?

process1 <- data.frame(x=vdata,subgroup=rep(1:25,each=5)) basechart <- ggplot(process1,aes(x=subgroup,y=x) ) xbar1 <- basechart + stat_QC(method = "xBar.rBar", auto.label = T, label.digits = 2)

Warning messages:
1: fun.y is deprecated. Use fun instead.
2: fun.y is deprecated. Use fun instead.

Thank you for your great work.

Hello,

I was attempting to extract my control limits and utilize those limits in additional plots in a sustainable manner so I wanted to extract a table format of the QC constants. However, upon doing so I found that the LCL and UCL calculated by the graph(Stat_QC) were different than the limits calculated by QC_Lines(). Is there a difference on how these are calculated?

If the dataset contains a "large" autocorrelation the control limits might not describe the natural process limits [1]. Thus, we need to be able to scale the lower and upper control limits by using an external/user-defined scaling factor. This scaling factor should have the default value 1, so that by default no scaling is active.

Would it be a major issue to include this option?

Reference:
[1] Donald Wheeler: "Advanced Topics in Statistical Process Control", SPC Press, 1995; Chapter 12 "Control Charts for Autocorrelated Data".

Observed that when n > 100 I got a convergence error for D5. Unfortunately this causes problems with all the metrics because this one fails.