knime/README.md

In jbaxx/ccharter: Control Charts made easy

Integration with KNIME: ccharter

I'll show you how the ccharter function can be integrated in a KNIME workflow. This is a work in progress, currently will require a bit of tweaking to make it work with your dataset.

We'll be using use the nodes in the picture below for this example.

1. Ensure you have installed the ccharter package in R and that KNIME uses your own R installation

   • Package installation instructions
   • Details for using R installation with KNIME are out of the scope of this guide

2. In a new KNIME workflow, create an R Source (Table) node. Press F6 to configure the node.

3. This node let's you read data from many sources, we pass the data read to the output port assigning it to a variable named knime.out. For this example we'll use a random generated dataset. You may copy/paste the code in the node.
   • Since R in KNIME has some issues passing date values, we'll always pass dates as characters to the output port.

  set.seed(154)

  time.series <- data.frame(t.dates = seq.Date(as.Date("2014-02-01"), as.Date("2016-08-01"), "month"),
                 t.values = c(
                 seq(0.1, 0.8, by = 0.1) * runif(8) + 3,
                 seq(0.1, 0.7, by = 0.1) * runif(7) + 4,
                 seq(0.1, 0.7, by = 0.1) * runif(7) + 5,
                 seq(0.1, 0.4, by = 0.1) * runif(4) + 4,
                 seq(0.1, 0.5, by = 0.1) * runif(5) + 4)
                 )
  time.series$t.dates <- as.character(time.series$t.dates)

  knime.out <- time.series # assign your data frame here

1. Create an R Snippet node:

• Here we call the ccharter library
• Execute the ccpoints function to calculate the Control Systems
• Extract the resulting data frame from the ccpoints object
• Send the data frame to the output port

library(ccharter)

knime.in$"t.dates" <- as.Date(knime.in$"t.dates")

control.chart.data <- ccpoints(knime.in, "t.dates", "t.values")
control.chart.data[["data"]]$t.dates <- as.character(control.chart.data[["data"]]$t.dates)

knime.out <- control.chart.data[["data"]]

1. Create an R View (Table) node:

• With this node we're going to plot the results
• Since we're not getting the ccpoints object from the previous node, we can't use the cc2plot function
• As workaround copy/paste the chunk of code below, it will plot the data frame received.

library(ggplot2)
library(scales)

knime.in$"t.dates" <- as.Date(knime.in$"t.dates")

  g <- ggplot(knime.in, aes(x = t.dates))
  g <- g + geom_line(aes(y = data.mean), size = 1.2, color = "darkorange")
  g <- g + geom_line(aes(y = data.ll), size = 1.2, color = "steelblue")
  g <- g + geom_line(aes(y = data.ul), size = 1.2, color = "steelblue")
  g <- g + geom_line(aes(y = t.values), size=0.6, color = "gray44")
  g <- g + geom_point(aes(y = t.values), color = "midnightblue")
  g <- g + theme_bw()
  g <- g + scale_x_date(labels = date_format("%b/%y"), minor_breaks = NULL, breaks = date_breaks("month"))
  g <- g + theme(axis.text.x = element_text(angle = 90, hjust = 1))
  g <- g + labs(x = "Month")
  g <- g + theme(plot.title = element_text(hjust = 0))
  print(g)

1. To view the plot, right-click the R View (Table) node and select View R: View\_:

1. Your result may look like this:

1. You may need to change the png settings in the R View (Table) node to set the desired dimensions

jbaxx/ccharter documentation built on Feb. 27, 2021, 9:10 p.m.