Cloudera Data Science Workbench demos

Basic tour of Cloudera Data Science Workbench.

Workbench

There are 4 scripts provided which walk through the interactive capabilities of Cloudera Data Science Workbench.

1. Basic Python visualizations (Python 2). Demonstrates:

• Markdown via comments
• Jupyter-compatible visualizations
• Simple console sharing

2. PySpark (Python 2). Demonstrates:

• Easy connectivity to (kerberized) Spark in YARN client mode.
• Access to Hadoop HDFS CLI (e.g. hdfs dfs -ls /).

3. Tensorflow (Python 2). Demonstrates:

• Ability to install and use custom packages (e.g. pip search tensorflow)

4. R Demonstrates:

• Run R code on CDSW, showing arules library

5. Advanced visualization with Shiny (R) Demonstrates:

• Use of 'shiny' to provide interactive graphics inside CDSW

Jobs

We recommend setting up a "Nightly Analysis" job to illustrate how data scientists can easily automate their projects.

Setup instructions

Note: You only need to do this once.

1. In a Python 3 Session:

!pip3 install --upgrade dask 
!pip3 install --upgrade keras 
!pip3 install --upgrade matplotlib==2.0.0. 
!pip3 install --upgrade pandas_highcharts 
!pip3 install --upgrade protobuf 
!pip3 install --upgrade tensorflow==1.3.0.
!pip3 install --upgrade seaborn
!pip3 install --upgrade numpy

Note, you must then stop the session and start a new Python session in order for all the packages to be seen.

2. In an R Session:

install.packages('sparklyr')
install.packages('plotly')
install.packages("nycflights13")
install.packages("Lahman")
install.packages("mgcv")
install.packages('shiny') 
install.packages("arules")
install.packages("readr")

3. Stop all sessions, then proceed.

‹