Next steps

Network analysis

A social graph of package Co-authors using tidyverse plus ggraph, an extension for ggplot2 for graphs (aka networks) and igraph, an extremely powerful network analysis library. Find the code and additional explanation here.

Semantic networks - www.swow.me

Text analysis

Another area were R has seen much development is text analysis. New packages provide access to lightning fast C++ string libraries, 'new' data formats such as html, xml, and json, and also to the API of social platforms, such as facebook.com and twitter.com.

Text analysis: Applications

Text analysis: Applications - hedonometer.org

Link to Paper
View Vonnegut's Shape of Stories

Stats packages

"If there is a new statistical method, it is first implemented in R"
-- duw

Deep learning

Deep learning: Tensorflow

# Get Keras
install.packages(keras)
library(keras)
install_keras()
# Define model
model <- keras_model_sequential() 
model %>% 
  layer_dense(units = 256,
              activation = 'relu', 
              input_shape = c(784)) %>% 
  layer_dropout(rate = 0.4) %>% 
  layer_dense(units = 128, 
              activation = 'relu') %>%
  layer_dropout(rate = 0.3) %>%
  layer_dense(units = 10, 
              activation = 'softmax')

Bayesian statistics

# load package
library(BayesFactor)
# define data
x <- rnorm(100, 0, 1)
y <- rnorm(100, 1, 1)
# run Frequentist analysis
t.test(x, y)
# run Bayesian analysis
ttestBF(x, y)

## Warning: package 'Matrix' was built under R version 3.4.4

## Bayes factor analysis
## --------------
## [1] Alt., r=0.707 : 5914249 ±0%
## 
## Against denominator:
##   Null, mu1-mu2 = 0 
## ---
## Bayes factor type: BFindepSample, JZS

Rcpp

By now one of the most referenced R packages is Rcpp - R's interface to C++. With often relatively little effort due to Rcpp sugar, Rcpp can provide vast speed improvements, which many packages today rely on Rcpp in the background for swift code execution. Rcpp becomes particularly powerful, when supplemented with BH, which makes avaialble a collection free, peer-reviewed C++ libraries, and RcppArmdillo, which available the high-performance Armadillo library for linear algebra methods.

Github & Dropbox

An important part of programming is to keep track of the progress to potential be able to go back to an earlier version of the code. To achieve this, many programmers rely on version-control software such as GitHub. R and, in particular, RStudio work have inbuilt software that allow you to sync your project with an GitHub repository. Via the package called rdrop2, R also interfaces with the popular cloud service Dropbox, which also allows for rudimentary version control.

R on servers

With the need for more computing power for, e.g., machine learning, R is also moving more and more to the server and cluster sphere. Rstudio offers a convenient (and free) solution to access RStudio sessions on a remote linux server via a browser. RStdudio also offers the sparklyr package integrating the Spark architecture for efficient cluster computing with dplyr. Other packages exist to interface with Hadoop (RHadoop) and SQL (RSQLite).

How to continue

Books

Here is a very incomplete series of good books. They are ordered by complexity, beginning with user-friendly books on learning statistics in R and ending with books focusing on the more advanced topics of the R language.

R meetup groups

Great place to meet other R enthusiast in and outside of industry and see interesting presentions on R's latest developments.

Consulting

Feel free to contact us.