ggedit – interactive ggplot aesthetic and theme editor

Guest post by Jonathan Sidi, Metrum Research Group

ggplot2 has become the standard of plotting in R for many users. New users, however, may find the learning curve steep at first, and more experienced users may find it challenging to keep track of all the options (especially in the theme!).

ggedit is a package that helps users bridge the gap between making a plot and getting all of those pesky plot aesthetics just right, all while keeping everything portable for further research and collaboration.

ggedit is powered by a Shiny gadget where the user inputs a ggplot plot object or a list of ggplot objects. You can run ggedit directly from the console from the Addin menu within RStudio.

Continue reading “ggedit – interactive ggplot aesthetic and theme editor”

Using 2D Contour Plots within {ggplot2} to Visualize Relationships between Three Variables

Guest post by John Bellettiere, Vincent Berardi, Santiago Estrada

The Goal

To visually explore relations between two related variables and an outcome using contour plots. We use the contour function in Base R to produce contour plots that are well-suited for initial investigations into three dimensional data. We then develop visualizations using ggplot2 to gain more control over the graphical output. We also describe several data transformations needed to accomplish this visual exploration.

Continue reading “Using 2D Contour Plots within {ggplot2} to Visualize Relationships between Three Variables”

heatmaply: interactive heat maps (with R)

I am pleased to announce heatmaply, my new R package for generating interactive heat maps, based on the plotly R package.


By running the following 3 lines of code:

heatmaply(mtcars, k_col = 2, k_row = 3) %>% layout(margin = list(l = 130, b = 40))

You will get this output in your browser (or RStudio console):

Continue reading “heatmaply: interactive heat maps (with R)”

Multidimensional Scaling with R (from “Mastering Data Analysis with R”)

Guest post by Gergely Daróczi. If you like this content, you can buy the full 396 paged e-book for 5 USD until January 8, 2016 as part of Packt’s “$5 Skill Up Campaign” at

Feature extraction tends to be one of the most important steps in machine learning and data science projects, so I decided to republish a related short section from my intermediate book on how to analyze data with R. The 9th chapter is dedicated to traditional dimension reduction methods, such as Principal Component Analysis, Factor Analysis and Multidimensional Scaling — from which the below introductory examples will focus on that latter.

Multidimensional Scaling (MDS) is a multivariate statistical technique first used in geography. The main goal of MDS it is to plot multivariate data points in two dimensions, thus revealing the structure of the dataset by visualizing the relative distance of the observations. Multidimensional scaling is used in diverse fields such as attitude study in psychology, sociology or market research.

Although the MASS package provides non-metric methods via the isoMDS function, we will now concentrate on the classical, metric MDS, which is available by calling the cmdscale function bundled with the stats package. Both types of MDS take a distance matrix as the main argument, which can be created from any numeric tabular data by the dist function.

But before such more complex examples, let’s see what MDS can offer for us while working with an already existing distance matrix, like the built-in eurodist dataset:

> as.matrix(eurodist)[1:5, 1:5]
          Athens Barcelona Brussels Calais Cherbourg
Athens         0      3313     2963   3175      3339
Barcelona   3313         0     1318   1326      1294
Brussels    2963      1318        0    204       583
Calais      3175      1326      204      0       460
Cherbourg   3339      1294      583    460         0

The above subset (first 5-5 values) of the distance matrix represents the travel distance between 21 European cities in kilometers. Running classical MDS on this example returns:

> (mds <- cmdscale(eurodist))
                      [,1]      [,2]
Athens           2290.2747  1798.803
Barcelona        -825.3828   546.811
Brussels           59.1833  -367.081
Calais            -82.8460  -429.915
Cherbourg        -352.4994  -290.908
Cologne           293.6896  -405.312
Copenhagen        681.9315 -1108.645
Geneva             -9.4234   240.406
Gibraltar       -2048.4491   642.459
Hamburg           561.1090  -773.369
Hook of Holland   164.9218  -549.367
Lisbon          -1935.0408    49.125
Lyons            -226.4232   187.088
Madrid          -1423.3537   305.875
Marseilles       -299.4987   388.807
Milan             260.8780   416.674
Munich            587.6757    81.182
Paris            -156.8363  -211.139
Rome              709.4133  1109.367
Stockholm         839.4459 -1836.791
Vienna            911.2305   205.930

These scores are very similar to two principal components (discussed in the previous, Principal Component Analysis section), such as running prcomp(eurodist)$x[, 1:2]. As a matter of fact, PCA can be considered as the most basic MDS solution.

Anyway, we have just transformed (reduced) the 21-dimensional space into 2 dimensions, which can be plotted very easily — unlike the original distance matrix with 21 rows and 21 columns:

> plot(mds)


Does it ring a bell? If not yet, the below image might be more helpful, where the following two lines of code also renders the city names instead of showing anonymous points:

> plot(mds, type = 'n')
> text(mds[, 1], mds[, 2], labels(eurodist))

Continue reading “Multidimensional Scaling with R (from “Mastering Data Analysis with R”)”

Slides from my JSM 2015 talk on dendextend

If you happen to be at the JSM 2015 conference this week, then this Monday, at 2pm, I will give a talk on the dendextend R package  (in the session “Advances in Graphical Frameworks and Methods Part 1“) – feel free to drop by and say hi.

Here are my slides for the intended talk:


p.s.: Yes – this presentation is very similar, although not identical, to the one I gave at useR2015. For example, I mention the new bioinformatics paper on dendextend.

dendextend: a package for visualizing, adjusting, and comparing dendrograms (based on a paper from “bioinformatics”)

This post on the dendextend package is based on my recent paper from the journal bioinformatics (a link to a stable DOI). The paper was published just last week, and since it is released as CC-BY, I am permitted (and delighted) to republish it here in full:


Summary: dendextend is an R package for creating and comparing visually appealing tree diagrams. dendextend provides utility functions for manipulating dendrogram objects (their color, shape, and content) as well as several advanced methods for comparing trees to one another (both statistically and visually). As such, dendextend offers a flexible framework for enhancing R’s rich ecosystem of packages for performing hierarchical clustering of items.

Availability: The dendextend R package (including detailed introductory vignettes) is available under the GPL-2 Open Source license and is freely available to download from CRAN at: (

Contact: [email protected]

Continue reading “dendextend: a package for visualizing, adjusting, and comparing dendrograms (based on a paper from “bioinformatics”)”

dendextend version 1.0.1 + useR!2015 presentation

When using the dendextend package in your work, please cite it using:

Tal Galili (2015). dendextend: an R package for visualizing, adjusting, and comparing trees of hierarchical clustering. Bioinformatics. doi:10.1093/bioinformatics/btv428

My R package dendextend (version 1.0.1) is now on CRAN!

The dendextend package Offers a set of functions for extending dendrogram objects in R, letting you visualize and compare trees of hierarchical clusterings. With it you can (1) Adjust a tree’s graphical parameters – the color, size, type, etc of its branches, nodes and labels. (2) Visually and statistically compare different dendrograms to one another.

The previous release of dendextend (0.18.3) was half a year ago, and this version includes many new features and functions.

To help you discover how dendextend can solve your dendrogram/hierarchical-clustering issues, you may consult one of the following vignettes:

Here is an example figure from the first vignette (analyzing the Iris dataset)



This week, at useR!2015, I will give a talk on the package. This will offer a quick example, and a step-by-step example of some of the most basic/useful functions of the package. Here are the slides:


Lastly, I would like to mention the new d3heatmap package for interactive heat maps. This package is by Joe Cheng from Rstudio, and integrates well with dendrograms in general and dendextend in particular (thanks to some lovely github-commit-discussion between Joe and I). You are invited to see lively examples of the package in the post at the RStudio blog. Here is just one quick example:

d3heatmap(nba_players, colors = “Blues”, scale = “col”, dendrogram = “row”, k_row = 3)


Israel’s 2015 election polls’ analysis with Shiny + ggplot2

(This is a guest post by my friend Yoni Sidi, a PhD candidate in statistics at the Hebrew University)


The Israeli elections are coming up this Tuesday, 17/3/2015 (i.e.: tomorrow!). They are a bit more complicated than your average US presidential race. The elections in Israel are based on nationwide proportional representation. The electoral threshold is 3.25% and the number of seats (or mandates) out of a total of 120 is proportional to the number of votes it recieves, so the threshold roughly translates to at least four mandates. The Israeli system is a multi-party system and is based on coalition governments. Multi-party is putting it mildly, there are 11 that have a chance (and are expected) to pass the mandate threshold.
There are two major parties, Hamachane Hazioni (Left Wing) and the Likud (Right Wing), that are hoping to garner between 16%-25% of the votes, 20-30 mandates. The main winners though are the medium size parties that recomend to the President who they think has the best chance to construct the next government, so yes there is a good possibility that the general elections winner will not be one constructing the coalition. Making the actual winners the parties that create the biggest coalition which exceeds 60 mandates.
An abundance of polling has been continually published during the run up and the variaety of pollsters and publishers is hard to keep track of as a casual voter trying to gauge the temperature of the political landscape. I came across a great realtime database by Project 61 on google docs of past and present polling result information and decided that it was a great opportunity to learn the Shiny library of RStudio and create an app that I can check current and past results. So after I figured out how to connect google docs to R, I created a self updating app that became a nice place to keep track of polling every day, check trends and distributions using interactive ggplot2 graphs and simulate coalition outcomes.
Please note that as of Friday (March 13th), until election day (March 17th), it is forbidden to perform new polls in Israel, hence the data presented here cannot allow for an up-to-date inference about the expected results of the election. This post is for educational purposes.

Running the election polls Shiny app on your computer

The github repo is available here.

#changing locale to run on Windows
if ([1] == "Windows") Sys.setlocale("LC_ALL","Hebrew_Israel.1255") 
#check to see if libraries need to be installed
libs <- c("shiny","shinyAce","httr","XML","stringr","ggplot2","scales","plyr","reshape2","dplyr")
x <- sapply(libs,function(x)if(!require(x,character.only = T)) install.packages(x))
#run App
#reset to original locale on Windows
if ([1] == "Windows") Sys.setlocale("LC_ALL")


Usage Instructions:

  1. Current Polling
  2. Election Analyis
  3. Mandate Simulator and Coalition Whiteboard
  4. Polling Database

Current Polling

  • The latest polling day results published in the media and the prediction made using the Project 61 weighting schemes. The parties are stacked into blocks to see which block has best chance to create a coalition.


The Project 61 prediction is based past pollster error deriving weights from the 2003,2006,2009 and 2013 elections, dependant on days to elections and parties. In their site there is an extensive analysis on pollster bias towards certain parties and party blocks.

Election Analysis

  • An interactive polling analysis layout where the user can filter elections, parties, publishers and pollster, dates and create different types of plots using any variable as the x and y axis.
  • The default layer is the 60 day trend (estimated with loess smoother) of mandates published by each pollster by party


The user can choose to include in the plots Elections (2003,2006,2009,2013,2015) and the subsequent filters are populated with the relevant parties, pollsters and publishers relevant to the chosen elections. Next there is a slider to choose the days before the election you want to view in the plot. This was used instead of a calendar to make a uniform timeline when comparing across elections.

In addition the plot itself is a ggplot thus the options above the graph give the user control on nearly all the options to build a plot. The user can choose from the following variables:

DaysLeftIdeology (5 Party Blocks)Mandate.GroupPollster
DateIdeology.Group (2 Party Blocks)Results
yearAttribute (Party History)(Pollster) Error

To define the following plot attributes:

Plot TypeAxesGroupingPlot Facets
PointX axis variableSplit Y by colors using a different variableRow Facet
BarDiscrete/ContinuousColumn Facet
LineRotation of X tick labels
StepY axis variable
  • Create Facets to display subsets of the data in different panels (two more variables to cut data) there are two type of facets to choose from
    • Wrap: Wrap 1d ribbon of panels into 2d
    • Grid: Layout panels in a grid (matrix)

An example of filtering pollsters to compare different tendencies for each party in the 2015 elections:


An example of comparing distribution mandates per party in the last two months of polling


An example of comparing distribution of pollster errors across elections (up to 10 days prior end of polling), by splitting the parties into five groups compared to previous election: old party,new party, combined (combination of two or more old parties), new.split (new party created from a split of a party from last election), old.split (old party that was a left from the split).



As we can see the pollster do not get a good indication of new,new.split or combined parties, which could be a problem this election since there are: 3 combined, 2 new splits.


  • If you are an R user and know ggplot there is an additional editor console,below the plot, where you can create advanced plots freehand, just add to the final object from the GUI called p and the data.frame is x, eg p+geom_point(). Just notice that all aesthetics must be given they are not defined in the original ggplot() definition. It is also possible to use any library you want just add it to the top of the code, the end object must be a ggplot.



#new layer
  • You can also remove the original layer if you want using the function remove_geom(ggplot_object,geom_layer), eg p=p+remove_geom(p,“point”) will remove the geom_point layer in the original graph


p=remove_geom(p,"point") #blank ggplot with facets in place
#new layer
  • Finally the plots can be viewed in English or Hebrew, and can be downloaded to you local computer using the download button.

Mandate Simulator and Coalition Whiteboard

  • A bootstrap simulation is run on Polling results from up to 10 of the latest polls using the sampling error as the uncertainty of each mandate published. Taking into account mandate surplus agreements using the Hagenbach-Bischoff quota method and the mandate threshold limit (in this election it is 4 mandates), calculating the simulated final tally of mandates. The distributions are plotted per party and the location of the median published results in the media.
  • The user can choose how many polls to take into account, up to last 10 polls, and how big a simulation they want to run: 50,100,500,1000 random polling results per each party and poll.


  • Once the simulator is complete you can create coalitions based on either the simulated distribution or actual published polls and see who can pass 60 mandates. Choose the coalition parties and the opposition parties from dropdown lists. (Yes the ones chosen are nonsensical on purpose…)


Polling Database

  • All raw data used in the application can be viewed and filtered in a datatable.

The dendextend package for visualizing and comparing trees of hierarchical clusterings (slides from useR!2014)

When using the dendextend package in your work, please cite it using:

Tal Galili (2015). dendextend: an R package for visualizing, adjusting, and comparing trees of hierarchical clustering. Bioinformatics. doi:10.1093/bioinformatics/btv428

This week I presented in the useR!2014 my package dendextend (also on github), for easily manipulating, visualizing, and comparing dendrograms. Put simply, it is a package designed to easily create figures like these:

2015-06-28 20_58_26-Clipboard

Here is my presentation from useR:

Download (PDF, Unknown)

You are also invited to give a look to the current version of the package vignettes:

I highly welcome features suggestions and bug reports (or just “wow, this is awesome”) sent to my e-mail (tal.galili AT, you can also leave a comment or use the github issue page.

A sidenote on useR!2014: this year’s useR conference was wonderful! I enjoyed the many talks, sessions, posters, and especially the so many wonderful R users I got to meet (and I will not try to list all of you – but you know who you are, and how much I enjoyed seeing you!). As corny as it may sound – we, the people who use R, are truly a community. There is a lot to be said about getting to meet so many people who share my own passion for statistical programming, open source, collaboration, open science, and a better future in general. Gladly, you can get a sense of what happened there by having a look at the twitter hashtag #useR2014. Several great R bloggers already started writing about it, you can see their posts here: 1, 2, 3, 4, 5. And I hope more posts will follow. I hope to see you in next year’s useR!2015!

Creating good looking survival curves – the 'ggsurv' function

This is a guest post by Edwin Thoen

Currently I am doing my master thesis on multi-state models. Survival analysis was my favourite course in the masters program, partly because of the great survival package which is maintained by Terry Therneau. The only thing I am not so keen on are the default plots created by this package, by using plot.survfit. Although the plots are very easy to produce, they are not that attractive (as are most R default plots) and legends has to be added manually. I come across them all the time in the literature and wondered whether there was a better way to display survival. Since I was getting the grips of ggplot2 recently I decided to write my own function, with the same functionality as plot.survfitbut with a result that is much better looking. I stuck to the defaults of plot.survfit as much as possible, for instance by default plotting confidence intervals for single-stratum survival curves, but not for multi-stratum curves. Below you’ll find the code of the ggsurv function. Just as plot.survfit it only requires a fitted survival object to produce a default plot. We’ll use the lung data set from the survival package for illustration. First we load in the function to the console (see at the end of this post).

Once the function is loaded, we can get going, we use the lung data set from the survival package for illustration.

lung.surv < - survfit(Surv(time,status) ~ 1, data = lung)


Continue reading “Creating good looking survival curves – the 'ggsurv' function”