H2o encoders starter

January 26, 2016

(This article was first published on koaning.io, and kindly contributed to R-bloggers)

This document contains a startup script for H2O in R. It is a silly example (why would anybody want to train a deep encoder on the iris dataset) but it helps people get started. This setup is meant for local use, not for cluster setup.

Just copy the code in!



h2o.init(ip = 'localhost', port = 54321, nthreads= -1, max_mem_size = '4g')


h2o_df <- as.h2o(iris)

mod_nn <- h2o.deeplearning(
  x = c('Sepal.Length', 'Sepal.Width', 'Petal.Length', 'Petal.Width'),
  training_frame = h2o_df,
  hidden = c(2),
  epochs = 100,
  activation = 'Tanh',
  autoencoder = TRUE

features <- h2o.deepfeatures(mod_nn, h2o_df, layer=1)
pltr_nn <- features %>% 
  as.data.frame %>% 
  mutate(species = iris$Species)

colnames(pltr_nn) <- c("x1", "x2", "species")

ggplot() + 
  geom_point(data=pltr_nn, aes(x1, x2, colour = species))


mod_pca <- h2o.prcomp(
  x = c('Sepal.Length', 'Sepal.Width', 'Petal.Length', 'Petal.Width'),
  training_frame = h2o_df, k = 2, transform = 'STANDARDIZE'

pltr_pca <- h2o.predict(mod_pca, h2o_df, num_pc=2) %>% 
  as.data.frame %>% 
  mutate(species = iris$Species)

colnames(pltr_pca) <- c("x1", "x2", "species")

ggplot() + 
  geom_point(data=pltr_pca, aes(x1, x2, colour = species))

The syntax is very lovely R-like. One just has to remember that h2o assumes a slightly different internal datastructure to keep things performant (similarly to SparkR). It does offer some machine learning possiblilities not available to SparkR users. Features!

You should also be able to find this blog on: r-bloggers

To leave a comment for the author, please follow the link and comment on their blog: koaning.io.

R-bloggers.com offers daily e-mail updates about R news and tutorials on topics such as: Data science, Big Data, R jobs, visualization (ggplot2, Boxplots, maps, animation), programming (RStudio, Sweave, LaTeX, SQL, Eclipse, git, hadoop, Web Scraping) statistics (regression, PCA, time series, trading) and more...

If you got this far, why not subscribe for updates from the site? Choose your flavor: e-mail, twitter, RSS, or facebook...

Comments are closed.

Search R-bloggers


Never miss an update!
Subscribe to R-bloggers to receive
e-mails with the latest R posts.
(You will not see this message again.)

Click here to close (This popup will not appear again)