topic Re: R Control Chart in Developer

R Control Chart

tglah — Wed, 08 Mar 2017 21:16:30 GMT

HelloPBI Community,

I havent found too much info on R Control Charts in this forum and am wondering if there is anyone out there that could copy&paste one?

Cheers

Re: R Control Chart

TedPattison — Wed, 08 Mar 2017 23:34:52 GMT

Here's is an example of creating a histogram with a density distribution line.

hist(dataset$Age, 
     main = "Customer Count by Age",
     ylab="Customer Count", 
     xlab="Customer Age",
     xlim = c(18, 100),
     border="black",
     breaks=20,
     col=c("lightyellow", "lightblue"), 
     las=1,
     probability = TRUE
)

lines(density(dataset$Age),lty="dotdash", lwd=4, col="red")

Here's an example of creating a barblot

barplot(dataset$'Sales Revenue',        
        names.arg = dataset$'Age Group', 
        main = "Sales Revenue by Customer Age Group",
        col = c("red","yellow","orange","blue", "green")
)

minValue <- 0
maxValue <- max(as.vector(dataset$'Sales Revenue'))
yTicks <- seq(from=minValue, to = maxValue, length.out = 10)
yTicks <- pretty(yTicks)
yTickLabels <- paste("$",format(yTicks/1000, , big.mark=","), "K",sep="")
axis(2, at=yTicks, labels = yTickLabels, lty = 1, las=1, cex.axis=0.7 )

Of course, these are simple examples using the built-in R graphics functionality. You can also use a richer graphics package such as lattice or ggplot2 to create some really detailed charts and graphs.

Is this what you are looking for?

Re: R Control Chart

tglah — Thu, 09 Mar 2017 16:45:37 GMT

Thanks Ted, this is along the same lines, but I am interested more in an "xbar" chart and also looking for a simple R script to create a Pareto chart for my customers.

Thank you for the scripts for the Histogram and Bar Plot, I will also be adding this to my arsenal.

Re: R Control Chart

MawashiKid — Tue, 14 Mar 2017 20:33:09 GMT

RE: also looking for a simple R script to create a Pareto chart

In order to use any chart -that is part of a R Visualization package library - it's logical to first make sure it's installed.

install.packages('<yourPackage>')

Now considering the huge amount of R package visualization libraries, there may be a few ways to reproduce what you want. Note that I don't consider myself an R guru at this stage, still here's a couple of basic samples.

qcc library...

library(qcc)
    defect <- c(80, 27, 66, 94, 33)
    names(defect) <- c("price code", "schedule date", "supplier code", "contact num.", "part num.")
    pareto.chart(defect, ylab = "Error frequency", col=heat.colors(length(defect)))

ggplot2 library...

library(ggplot2)

counts  <- c(80, 27, 66, 94, 33)
defects <- c("price code", "schedule date", "supplier code", "contact num.", "part num.")

dat <- data.frame(
  count = counts,
  defect = defects,
  stringsAsFactors=FALSE
)

dat <- dat[order(dat$count, decreasing=TRUE), ]
dat$defect <- factor(dat$defect, levels=dat$defect)
dat$**bleep** <- cumsum(dat$count)
dat

ggplot(dat, aes(x=defect)) +
  geom_bar(aes(y=count), fill="blue", stat="identity") +
  geom_point(aes(y=**bleep**)) +
  geom_path(aes(y=**bleep**, group=1))

I haven't played much with x-Bar chart though I'll check if I can find something... Anyway hope this helps

Re: R Control Chart

MawashiKid — Thu, 16 Mar 2017 21:09:33 GMT

RE : xBarChart

In true life scenario, I'd say the following data would probably be obtained as a result
of fancy mathemathic algorithm formula... though I'll leave that to Math PhDs...
So for for simplicity I just added a list of subgroup manually...

So here we go...

#declare qcc library:

library(qcc)

# Load a mock list of 10 subgroup data manually:

sg1 <- c(1.397742,1.399917,1.278918,1.279828) # Fill in subgroup 1 data!
sg2 <- c(1.283877,1.307215,1.341566,1.396107) # Fill in subgroup 2 data!
sg3 <- c(1.313634,1.278839,1.242498,1.331201) # Fill in subgroup 3 data!
sg4 <- c(1.245943,1.303432,1.390168,1.298949) # Fill in subgroup 4 data!
sg5 <- c(1.188624,1.226905,1.217450,1.284127) # Fill in subgroup 5 data!
sg6 <- c(1.287875,1.283185,1.234186,1.314055) # Fill in subgroup 6 data!
sg7 <- c(1.276711,1.284376,1.305309,1.249184) # Fill in subgroup 7 data!
sg8 <- c(1.312219,1.297509,1.272367,1.371223) # Fill in subgroup 8 data!
sg9 <- c(1.378350,1.312981,1.381944,1.268875) # Fill in subgroup 9 data!
sg10 <- c(1.332196,1.268824,1.299608,1.329053) # Fill in subgroup 10 data!

# Include those subgroups into a my.data mock list through rbind

my.data <- rbind(sg1,sg2,sg3,sg4,sg5,sg6,sg7,sg8,sg9,sg10)

# Draw the R Chart and calculate relevant metrics

q1 <- qcc(my.data, type="R", nsigmas=3)

which should generate similar R chart:

then add following to ldraw the X-BAR chart...

# Draw the X-Bar Chart and calculate relevant metrics
q2 <- qcc(my.data, type="xbar", nsigmas=3)

which should generate following X-Bar Chart:

Finally:

# Establish the LSL and USL as set by customer specs, then
# draw the process capability chart and calculate metrics:

lsl <- 1.31 # Fill in a mock LSL here!
usl <- 1.32 # Fill in a mock USL here!

process.capability(q2, spec.limits=c(lsl,usl))

which should end up with following:

Hope this helps...

Re: R Control Chart

tglah — Fri, 17 Mar 2017 16:01:21 GMT

Thank you for all of this information! I am going to try and take some time this weekend to put into play your examples.

Much appreciated! Ill reply back with any progress I can make.

Cheers.

Re: R Control Chart

tglah — Mon, 20 Mar 2017 13:52:30 GMT

I have been able to create the R and XBar charts in PowerBI R Visuals, but still need some adjustments as it doesnt look right to me (data not displaying as expected).

library(qcc)

months<- c("Jan", "Feb", "Mar", "Apr","May", "Jun","Jul","Aug","Sep","Oct","Nov","Dec")

qcc(dataset, type="R", nsigmas=1, labels=months, xlab= "Month", ylab = "Service Level %", title = "Phone Call SVL", digits=3)

Any idea how to get the Y axis to show as percentages? My data is in percentages as you can see when selecting a simple Line Chart from the PowerBI Visuals.

Thanks for all your assistance!

Re: R Control Chart

MawashiKid — Tue, 28 Mar 2017 00:26:28 GMT

Sorry for late response. For the sake of simplicity, I mainly used mock data which I entered
manually. The idea was basically to create a series of 4 weeks subsets over a 12 months period sort if thing...
We both agree that a real life scenario would have gone through a much fancier scenario such as querying data
from a server and even adding conversion at some stage. I used numeric for simplicity.

Also I noticed that even if qcc library offers a wide variety of statisitic chart [R, xBar,...], it doesn't seem to have
any qcc - y axis formatting options parameter as you would probably find using scaling_y_continuous in ggplot2 library.
So in this case I don't believe y axis percentage formatting could be done in one shot.

My best bet would be to define the qcc : q1 R Chart and q2 xBar in respectice class with a plot=False attribute

 library(qcc)
 Jan <- c(0.837742,0.839917,0.728918,0.729828) # Fill in subgroup January data!
 Feb <- c(0.783877,0.807215,0.841566,0.836107) # Fill in subgroup February data!
 Mar <- c(0.813634,0.728839,0.742498,0.831201) # Fill in subgroup March data!
 Apr <- c(0.745943,0.803432,0.830168,0.798949) # Fill in subgroup April data!
 May <- c(0.688624,0.726905,0.717450,0.784127) # Fill in subgroup May data!
 Jun <- c(0.787875,0.783185,0.714186,0.814055) # Fill in subgroup June data!
 Jul <- c(0.726711,0.784376,0.805309,0.749184) # Fill in subgroup July data!
 Aug <- c(0.812219,0.797509,0.722367,0.871223) # Fill in subgroup August data!
 Sep <- c(0.878350,0.812981,0.881944,0.768875) # Fill in subgroup September data!
 Oct <- c(0.832196,0.768824,0.799608,0.729053) # Fill in subgroup October data!
 Nov <- c(0.813634,0.728839,0.742498,0.831201) # Fill in subgroup November data!
 Dec <- c(0.726711,0.784376,0.805309,0.749184) # Fill in subgroup December data!
 
 # Include those subgroups into a my.data mock list through rbind
 
 dataset <- rbind(Jan,Feb,Mar,Apr,May,Jun,Jul,Aug,Sep,Oct,Nov,Dec)
 months<- c("Jan", "Feb", "Mar", "Apr","May", "Jun","Jul","Aug","Sep","Oct","Nov","Dec") 

 # assign q1
 q1 <- qcc(dataset, type="R", nsigmas=1, labels=months, xlab= "Month", ylab = "Service Level %", title = "Phone Call SVL", digits=3, label.limits = c("5%", "14%"),plot=FALSE)
 # assign q2
 q2 <- qcc(dataset, type="xbar", nsigmas=1, labels=months, xlab= "Month", ylab = "Service Level %", title = "Phone Call SVL", digits=3, label.limits = c("76%", "80%"), plot=FALSE)

and then use each q1 & q2 with a plot method allowing to erase the y axis and redifine a new one with somekind of thicks.

plot(q1, yaxt="n", ... , xlab="Months", ylab="Service Levels %", title="Phone Call SVL", label.limits= c("5%", "13%"))

plot(q2, yaxt="n", ...,  xlab="Months", ylab="Service Levels %", title="Phone Call SVL", label.limits= c("76%", "80%"))

Tom Hopper has written a blog on rewriting qcc plot using ggplot2 and grid,
https://tomhopper.me/2014/03/03/rewriting-plot-qcc-using-ggplot2-and-grid/

though I haven't got a chance to dive deep into it. I wish I could bring you an easiy solution on y axis percentage formatting
though I haven't found one with qcc. Sorry.

Re: R Control Chart

makika — Mon, 23 Oct 2017 10:27:55 GMT

edited: learned how to refer to individual column in dataset and how to define x-axis labels. still, the sample order is wrong..

edited2: got it solved myself. details below to other learners like me.

Hello,

I've been trying to create xbar.one (individual) chart with power bi / qcc, with no luck so far. It seems values inside dataset somehow automatically get rearranged to ascending order. If I provide the values in script manually, chart looks fine. But if they come from Excel or SQL table (power bi's automatically formed dataset), values get rearranged. ~~I'd also like to bind the values to labels on x-axis, but haven't found the way how.~~

I'm new to R.

Samples:

360.1

360.2

360.0

359.8

Any help is greatly appreciated.

Solved:

newdata <- dataset[order(dataset$SERIALID),]

library(qcc)

x <- c(newdata$SERIALID)

s <- c(newdata$NUM_VALUE)

qcc(s, type="xbar.one", labels=x)

- kari