From Data to Models

This repository includes data and R scripts for the tutorial:

From Data to Models: classification, prediction, and synthesis

given by Carlo A. Furia at USI in October 2019.

All data is from public datasets; see files sources for links to the original sources.

Requirements to run the tutorial

1. Install the R platform from https://stat.ethz.ch/CRAN/ by following the instructions for your operating system

2. (Optional, but recommended) Install the RStudio IDE from https://www.rstudio.com/products/rstudio/download/ by choosing the free version of RStudio Desktop and following the instructions for your operating system

3. Get a snapshot of this repository:

   • Download the archive file https://github.com/bugcounting/data2models/archive/master.zip
   • Unpack it into a directory in your system (we'll call it root in these instructions)

4. Open RStudio and issue the following commands from the pull-down menus:

   • File -> Open File -> pick file root/data2models-master/spam/pull_data.R
   • Session -> Set Working Directory -> To Source File Location
   • Code -> Source

   This will download the remaining libraries and data

Credits and references

• The book Machine learning for hackers by Drew Conway and John Myles White (O'Reilly, 2012) is a practical presentation of several machine learning techniques (including naive Bayes classifiers and linear regression) with complete code examples in R. The example of Bayesian spam classifier is based on chapter 3 of the book.

• Dirk Schumacher, the author of the rpicosat library used in the Sudoku example, has a blog post where he describes how to build the same propositional Sudoku constraints encoding we use in sat/sat.R but using a different approach in R.