Discovering Data Science Design Patterns with Examples from R and Python Software Ecosystem

This repository contains source code & other data for my graduate thesis.

All pictures are located in tex_thesis/images folder.

These are just necessary for running code samples.

For Python 3, execute:

sudo apt-get install python3-pip r-base r-base-dev 
sudo pip3 install jupyter pandas scipy jupyterhub fancyimpute tensorflow networkx==1.11 scikit-learn numpy mlens
sudo pip3 install -r code_data/dp_9/python-plotly/requirenments.txt

For R, install packages below to cover most important ones -- others will be installed too. It will take a long time but is worth it.

ipak <- function(pkg){
new.pkg <- pkg[!(pkg %in% installed.packages()[, "Package"])]
if (length(new.pkg))
install.packages(new.pkg, dependencies = TRUE)
sapply(pkg, require, character.only = TRUE)
}
required.packages <- c("ggplot2", "tidyverse", "reshape2", "shiny", "caret", "knitr", "Rcpp", "rmarkdown", "pROC", "data.table", "mice", "mlbench", "caretEnsemble", "glmnet", "pls", "caTools")))
ipak(packages)

(tested on Ubuntu OS 17.10)

To reproduce thesis document, install full Tex with biblatex/biber and git-scm

sudo apt-get install texlive-full git

Once done (it take a very long time), one needs to have specific fonts installed as well. First, those from Adobe for the source code, execute:

[ -d /usr/share/fonts/opentype ] || sudo mkdir /usr/share/fonts/opentype
sudo git clone --depth 1 --branch release https://github.com/adobe-fonts/source-code-pro.git /usr/share/fonts/opentype/scp
sudo fc-cache -f -v
fc-cache -f -v

Moreover, one needs to have 4 fonts similar to EB(/Adobe) Garamond: Bold, BoldItalic, Regular and Italic (not the "cond" and "display" versions, just "classical"). You will not find them here due to licensing but make sure to put them into the custom_fonts folder and adjust thesis.tex file as needed.

In the code_data folder, execute sudo ./create_pdf_thesis.sh bash script. Based on the content of the tex_thesis folder and if fonts are properly installed, this will ultimately create PDF (~ 300 seconds/5 minutes is minimum what it takes to compile).

Source templates come from adjusting https://www.sharelatex.com/templates/thesis/lmu-thesis and mainly from https://www.sharelatex.com/templates/thesis/university-of-michigan-thesis

• For Plot.ly Dash:

  • Run python3 app.py in dp_9/python-plotly folder (https://designpattern10.herokuapp.com/ might be still working well)

• R(md) & Shiny: Use RStudio IDE to run dp_9/R-Shiny/app.R or .Rmd files.

  • Or Rscript -e "library(rmarkdown);rmarkdown::render('R_masterThesis.Rmd')" from bash.

• Jupyter Notebooks: open any browser and

  • Run jupyterhub --no-ssl from bash and navigate to desired ipynb file.
  • Alternatively, just run jupyter notebook.

https://stackoverflow.com/a/3452888

https://stackoverflow.com/a/15593865

https://gist.github.com/stevenworthington/3178163

https://github.com/jupyterhub/jupyterhub/wiki/Installation-of-Jupyterhub-on-remote-server

adobe-fonts/source-code-pro#17 (comment)