Using R for Advanced Charts When you need to build charts beyond Excel's capabilities

Should I Use R?               (See comparison of Excel & R Trellis Charts: video)

In pursuing my interests in data visualization, I have found several authors who have really had an impact on my graphical analysis thinking:

Edward Tufte - I first started my data visualization journey when I by read Tufte's Visual Display of Quantitative Information. While comfortable with Tufte's concepts, I found I needed more concrete examples and tools. This lead me to Naomi Robbins's book.

Naomi  Robbins' Creating More Effective Graphs book opened my eyes to a wide array of practical charting techniques not available in Excel. I was particularly interested in dot plots and trellis plots. Robbins' book is where I first learned about the R, S, and S Plus statistical and graphics languages.  Her chart examples and discussion lead me to William Cleveland's and Paul Murrell's books.

William Cleveland' s Elements of Graphing Data and  Visualizing Data books showed me that I was just scratching the surface of graphical analysis with default Excel charts. In addition to trellis displays, Cleveland's books showed me important concepts like banking to 45^o and locally weighted  regression (Lowess). 

Paul Murrrell's R Graphics  showed me the power of the R graphics package.

As I read Robbins, Cleveland and Murrell's books, I saw a number of data analysis and charting capabilities that I wanted to add to my repertoire:

• Trellis charts ( I started with horizontal and vertical panel charts in Excel)

• Box plots

• Dot plots

• Banking to 45^o - (Uses Excel Solver)

• Lowess ( No Excel progress)

While trying to make these advanced Excel based charts,  I often asked myself the question "...should I use R for this chart instead of  Excel". Based on Robbins' book, I knew that these charts were readily available in R.

Reluctant to take on the R learning curve, I developed several of the advanced charts in Excel and VBA. I knew it was taking more time than it would to use a high quality statistical analysis and graphing package. I felt that it was better to use a tool that I knew and avoid the learning curve of a package like R.

I made good progress building dot plots, box plots and banking to 45^o tools in Excel. Trellis charts  and Lowess smoothing proved a real challenge in Excel. Trellis charts allow the user to show multivariate data much more effectively than Excel's default charts. I looked into R again  and considered using R for my trellis charting needs. Rather than switch to R at that time, I started developing horizontal and vertical panel charts in Excel, wanting to be able to automate the development of a full trellis chart.

I searched high and low for a reasonable Lowess algorithm that I could incorporate into Excel. No luck.

I finally developed a set of VBA procedures to build a full trellis chart.  When I completed the Excel - VBA trellis chart, I reconsidered R because I now know what it will take for me to develop the trellis advanced charting that was already available in R.

Why reinvent trellis charts in Excel when they were readily available in R?  To answer this question, I needed to find out how difficult the R learning curve was, I already knew what it takes to develop these tools in Excel - VBA.

Trellis Chart in Excel and R

This video shows my Excel VBA trellis chart tool as well as the results of my 2 day R learning curve.

Based on my Excel - VBA trellis chart development time versus R trellis chart learning curve time, I will be using R for graphics that is not part of Excel's standard tools kit. I'd rather use a proven tool by a R programmer than reinvent my own graphic wheel in Excel.

I have a long way to go in mastering data analysis and time series analysis. I'd rather spend my time learning and using analysis  techniques rather than coaxing Excel to do what R can already do.

R Resources

(top)

R Lattice Beats Excel's Stacked Area Trend Chart

See Charts & Graphs blog for my post on why R lattice plots are much more effective than Excel's stacked area trend charts.

Here's the link to the source data file.

Here's the R script I used.

### Script to work with BP Oil's Oil Consumption - M Barrels/day Data
# D Kelly O'Day - Oct. 3, 2008; http://processtrends.com & http://chartsgraphs.wordpress.com
library(lattice)
my_ar <- c(1) # User changeable aspect ratio
# *** Please edit for correct path to your source data file
my_data <- read.table("c:/R_home/CG/oil_cons_trend.csv", sep =",", header=TRUE)
# *****
tp1 <- xyplot(Value ~ Year |Parameter, data = my_data, type = "l",
main="World Oil Consumption Trends - million barrels per day \n 1965 to 2007 ",
sub = "Data Source: BP Statistical Review of World Energy 2008",
xlab="",
ylab="Million barrels per day",
par.settings=list(axis.text=list(cex=0.8), fontsize=list(text=10)),
par.strip.text=list(cex=0.8))
update(tp1, aspect = my_ar)

(top)

See Charts & Graphs blog for my post on my walkthrough of an R script..

Here's the link to the R script file..

(top)

See Charts & Graphs blog for my post on how to make a step chart with R.

Here are the links to R script file and data file.

(top)

See my Charts & Graphs blog post on how to make a boxplot with R.

Here are the links to the R script and data file for boxplots.

(top)

Here are the links to the R script, ENSO and GISS temp data files for the trend chart with change points..

If you try the script, be sure to change the source data file path to the actual folder where you saved the data files.

(top)

See my Charts & Graphs blog post on how R handles categorical variables and hte power of factors in data analysis and plotting.

Here are the links to the R script  and Enso_GISS temp data file for the trend chart with change points..

If you try the script, be sure to change the source data file path to the actual folder where you saved the data files.

(top)

This R script shows the long term global land and sea temperature trends,  showing  monthly and annual anomalies, long term trend (lowess fit [(f = 0.15]) as well as the 2001-2008 trend. I prepared this chart using R. 

See my Charts & Graphs post here.

The R script file is here.

(top)

R script to automatically generate a trend chart from web based global temperature data.  The R script allows me to update my plot each month as soon as the source file is updated. The plot is self documenting so that the chart reader can see the data period,  the overall trend line, slope for trend line; the last data point is highlighted and value given. Date stamp and name are included in the margin.

See my Chart & Graphs post here.

R Script file link.

(top)

In this post, I show how to make a 3 variable time series panel chart in R. As an Excel panel chart pioneer, I can say that it would be very difficult and messy to produce a comparable chart in Excel.  The example R chart uses panel chart, step chart, error bar and dynamic capabilities not inherently available in Excel. No helper series are needed and the chart can easily be regenerated each month as new data is available.

See my Chart & Graphs post here.

Data file is here.

R script file is here.

(top)

In a previous post, I showed how to make a 3 panel chart displaying RSS temperature anomalies,  El Nino- La Nina (NINO34 index) and volcanic event (SATO Index) time series data.  While time series charts can give a visual impression of how time series are related, time series regression provides much more information, including leading - lagging relationships between the time series.

In this post, I analyze the RSS, NINO34 and SATO time series using a simple trend line and a multiple regression model with lag periods for the independent variables. The post demonstrates some of  R 's regression analysis and data lagging capabilities. Links to the R script and data files are provided.

See my Chart & Graphs post here.

Data file is here.

R script file is here.

(top)

  script              data

(top)

See my Chart & Graphs post here.

top)