Python

Materials for class on Monday, April 29, 2019

Contents

Tweet of the day

Just released Altair version 3.0! This is a huge update, with many new visualization features and bug-fixes thanks to the hard work of the ??? team.

Check it out: https://t.co/PCyrIOTcvv pic.twitter.com/0YT2ckQizl

— Jake VanderPlas (@jakevdp) April 26, 2019

Altair is based on Vega and Vega-Lite, another creation by Jeff Heer and his lab. You can try Vega and Vega-lite here! There is also a vegalite R package.

Lab: Python+R (reticulate) in a Docker Container

1. Run RStudio’s ML Docker Image in one of these ways:

Did you know, this docker image also includes Machine Learning packages (for CPU) like h2o, tensorflow, and keras! It can expand to GPU environments too!

1. Easiest option (free, no local install): play-with-docker

2. Will take time to install (free): local Docker

3. Will cost money: Digital Ocean Droplet

# run in bash command/terminal
docker run -e PASSWORD=[create_a_pwd] -p 8787:8787 rocker/ml

⊕

3. Open RStudio via port 8787

FYI: username: rstudio, pwd: whatever you chose

4. Importing python packages

# run in RStudio console/file
# fyi reticulate is prepackaged
library(reticulate)
os <- import("os")
os$listdir(".")

5. Installing python packages

1. Run this in RStudio terminal:

#run in rstudio terminal
easy_install pip
easy_install virtualenv

2. Restart RStudio session (Session > Restart R)

3. Run this in RStudio console:

#run in rstudio console
reticulate::py_install("pandas")

Altair via Jupyter Docker install (“no R”)

1. Altair Jupyter notebook installation

# see https://jupyter-docker-stacks.readthedocs.io/en/latest/
docker run -p 8888:8888 jupyter/minimal-notebook:17aba6048f44

2. Open the Jupyter notebook in port 8888

3. Open terminal via Jupyter and run:

conda install -c conda-forge altair vega_datasets notebook vega
git clone https://github.com/altair-viz/altair-tutorial.git

4. Navigate to the notebook “01-Cars-Demo.ipynb”

See the folder “/altair-tutorial/notebooks/”.

5. Comment this out in Jupyter notebook.

alt.renderers.enable('notebook')

Links

1. Docker training

2. Altair tutorial jupyter notebooks, slides and video

3. Yihui’s thoughts on R-Python notebooks

4. Python reticulate configuration

5. RStudio reticulate cheatsheet