by Simon Jackson

R tips and tricks from a scientist. All R Markdown docs with full R code can be found at my GitHub:

Read this first

Creating corporate colour palettes for ggplot2

@drsimonj here to share how I create and reuse corporate color palettes for ggplot2.

You’ve started work as a data scientist at “drsimonj Inc” (congratulations, by the way) and PR have asked that all your Figures use the corporate colours. They send you the image below (coincidentally the Metro UI colors on


You want to use these colours with ggplot2 while also making your code reusable and flexible.

 Outline and setup

We’re going to create the following:

  1. Named vector of hex codes for the corporate colors
  2. Function to access hex codes (in 1)
  3. Named list of corporate color palettes (combinations of colors via 2)
  4. Function to access palettes (in 3)
  5. ggplot2-compatible scale functions that use the corporate palettes (via 4)

Load the ggplot2 package and set a default theme to setup:



 Start with color

Everything starts

Continue reading →

Five tips to improve your R code

@drsimonj here with five simple tricks I find myself sharing all the time with fellow R users to improve their code!

This post was originally published on DataCamp’s community as one of their top 10 articles in 2017

 1. More fun to sequence from 1

Next time you use the colon operator to create a sequence from 1 like 1:n, try seq().

# Sequence a vector
x <- runif(10)
#>  [1]  1  2  3  4  5  6  7  8  9 10

# Sequence an integer
#>  [1]  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
#> [24] 24 25 26 27 28 29 30 31 32

The colon operator can produce unexpected results that can create all sorts of problems without you noticing! Take a look at what happens when you want to sequence the length of an empty vector:

# Empty vector
x <- c()

#> [1] 1 0

#> integer(0)

You’ll also notice that this saves you from using functions

Continue reading →

ggplot2 SEM models with tidygraph and ggraph

@drsimonj here to share a ggplot2-based function for plotting path analysis/structural equation models (SEM) fitted with Yves Rosseel’s lavaan package.


SEM and its related methods (path analysis, confirmatory factor analysis, etc.) can be visualized as Directed Acyclic Graphs with nodes representing variables (observed or latent), and edges representing the specified relationships between them. For this reason, we will use Thomas Lin Pedersen’s tidygraph and ggraph packages. These packages work together to work with relational structures in a tidy format and plot them using ggplot2.

 The function

Below is a function ggsem(), which takes a fitted lavaan object and returns a ggplot2 object representing the nodes, edges, and parameter values. It handles regression paths, correlations, latent factors, and factor loadings.


Continue reading →

Big Data Solutions: A/B t test

@drsimonj here to share my code for using Welch’s t-test to compare group means using summary statistics.


I’ve just started working with A/B tests that use big data. Where once I’d whimsically run t.test(), now my data won’t fit into memory!

I’m sharing my solution here in the hope that it might help others.

 In-memory data

As a baseline, let’s start with an in-memory case by comparing whether automatic and manual cars have different Miles Per Gallon ratings on average (using the mtcars data set).

t.test(mpg ~ am, data = mtcars)
#>  Welch Two Sample t-test
#> data:  mpg by am
#> t = -3.7671, df = 18.332, p-value = 0.001374
#> alternative hypothesis: true difference in means is not equal to 0
#> 95 percent confidence interval:
#>  -11.280194  -3.209684
#> sample estimates:
#> mean in group 0 mean in group 1 
#>        17.14737        24.39231

Well… that was easy!

Continue reading →

A tidy model pipeline with twidlr and broom

@drsimonj here to show you how to go from data in a data.frame to a tidy data.frame of model output by combining twidlr and broom in a single, tidy model pipeline.

 The problem

Different model functions take different types of inputs (data.frames, matrices, etc) and produce different types of output! Thus, we’re often confronted with the very untidy challenge presented in this Figure:


Thus, different models may need very different code.

However, it’s possible to create a consistent, tidy pipeline by combining the twidlr and broom packages. Let’s see how this works.

 Two-step modelling

To understand the solution, think of the problem as a two-step process, depicted in this Figure:


 Step 1: from data to fitted model

Step 1 must take data in a data.frame as input and return a fitted model object. twidlr exposes model functions that do just this!

To demonstrate:

Continue reading →

Pretty scatter plots with ggplot2

@drsimonj here to make pretty scatter plots of correlated variables with ggplot2!

We’ll learn how to create plots that look like this:



In a data.frame d, we’ll simulate two correlated variables a and b of length n:

n <- 200
d <- data.frame(a = rnorm(n))
d$b <- .4 * (d$a + rnorm(n))

#>            a           b
#> 1 -0.9279965 -0.03795339
#> 2  0.9133158  0.21116682
#> 3  1.4516084  0.69060249
#> 4  0.5264596  0.22471694
#> 5 -1.9412516 -1.70890512
#> 6  1.4198574  0.30805526

 Basic scatter plot

Using ggplot2, the basic scatter plot (with theme_minimal) is created via:


ggplot(d, aes(a, b)) +
  geom_point() +


 Shape and size

There are many ways to tweak the shape and size of the points. Here’s the combination I settled on for this post:

ggplot(d, aes(a, b)) +
  geom_point(shape = 16, size = 5) +

Continue reading →

Pretty histograms with ggplot2

@drsimonj here to make pretty histograms with ggplot2!

In this post you’ll learn how to create histograms like this:


 The data

Let’s simulate data for a continuous variable x in a data frame d:

d <- data.frame(x = rnorm(2000))

#>            x
#> 1  1.3681661
#> 2 -0.0452337
#> 3  0.0290572
#> 4 -0.8717429
#> 5  0.9565475
#> 6 -0.5521690

 Basic Histogram

Create the basic ggplot2 histogram via:


ggplot(d, aes(x)) +


 Adding Colour

Time to jazz it up with colour! The method I’ll present was motivated by my answer to this StackOverflow question.

We can add colour by exploiting the way that ggplot2 stacks colour for different groups. Specifically, we fill the bars with the same variable (x) but cut into multiple categories:

ggplot(d, aes(x, fill = cut(x, 100))) +


What the…

Oh, ggplot2 has

Continue reading →

twidlr: data.frame-based API for model and predict functons

@drsimonj here to introduce my latest tidy-modelling package for R, “twidlr”. twidlr wraps model and predict functions you already know and love with a consistent data.frame-based API!

All models wrapped by twidlr can be fit to data and used to make predictions as follows:


fit <- model(data, formula, ...)
predict(fit, data, ...)
  • data is a data.frame (or object that can be corced to one) and is required
  • formula describes the model to be fit

 The motivation

The APIs of model and predict functions in R are inconsistent and messy.

Some models like linear regression want a formula and data.frame:

lm(hp ~ ., mtcars)

Models like gradient-boosted decision trees want vectors and matrices:


y <- mtcars$hp
x <- as.matrix(mtcars[names(mtcars) != "hp"])

xgboost(x, y, nrounds = 5)

Models like generalized linear models want you to work. For

Continue reading →

How and when: ridge regression with glmnet

@drsimonj here to show you how to conduct ridge regression (linear regression with L2 regularization) in R using the glmnet package, and use simulations to demonstrate its relative advantages over ordinary least squares regression.

 Ridge regression

Ridge regression uses L2 regularisation to weight/penalise residuals when the parameters of a regression model are being learned. In the context of linear regression, it can be compared to Ordinary Least Square (OLS). OLS defines the function by which parameter estimates (intercepts and slopes) are calculated. It involves minimising the sum of squared residuals. L2 regularisation is a small addition to the OLS function that weights residuals in a particular way to make the parameters more stable. The outcome is typically a model that fits the training data less well than OLS but generalises better because it is less sensitive to extreme

Continue reading →

Easy leave-one-out cross validation with pipelearner

@drsimonj here to show you how to do leave-one-out cross validation using pipelearner.

 Leave-one-out cross validation

Leave-one-out is a type of cross validation whereby the following is done for each observation in the data:

  • Run model on all other observations
  • Use model to predict value for observation

This means that a model is fitted, and a predicted is made n times where n is the number of observations in your data.

 Leave-one-out in pipelearner

pipelearner is a package for streamlining machine learning pipelines, including cross validation. If you’re new to it, check out blogR for other relevant posts.

To demonstrate, let’s use regression to predict horsepower (hp) with all other variables in the mtcars data set. Set this up in pipelearner as follows:


pl <- pipelearner(mtcars, lm, hp ~ .)

How cross validation is done is handled by learn_cvpairs()

Continue reading →