Analysis scripts for the watchdog MongoDB database

In order to run the R analyses in this repository, you must complete the following steps, detailed later in the Readme.

1. Required: All collections are in a local or remote mongo db
2. Install all R dependencies by running packages.R
3. Create config.R that contains connection information to mongodb*
4. Execute generate scripts, generate (in this order) users.csv, projects.csv, intervals.csv and sequences.csv
5. Run research analysis: Execute script batch_start.R

Install R dependencies

To initialize R with the required library dependencies, do the following:

cd R
R --no-save < src/packages.R

Setup Mongo connection for R

You setup the conneciton to your Mongo databse via files R/config.R and R/passwords.R.

mongo.user    <- "watchdog"
mongo.passwd  <- "watchdog"
mongo.host    <- "127.0.0.1"
num.processes <- 2
base.dir      <-  "."

If you do not run MongoDB locally, you can open an ssh tunnel to dutiap.st.ewi.tudelft.nl as follows:

ssh -L 27017:dutiap:27017 dutiap

Execute generate scripts

We will first need to generate the CSV files that our R analysis can read-in from a Mongo database. You need to setup the connection to the mongo database (see previous step). This can either be local or remote. The generation code is meant to be run with Rscript as follows:

Rscript src/generate-watchdog-csv.R [options] cmd

At the moment, the following cmds exist (run them in this order)

• gen-users-file: Create a CSV file with all registered users.
• gen-projects-file: Create a CSV file with all registered projects.
• gen-intervals-file: Create a CSV file with all intervals flattened.
• gen-sequence-file: Create a CSV file with all intervals sequentialized for TDD analysis.
• gen-reports (optional): Generate test reports for all projects

The following options exist: -n x: The number of processors to use for this analysis (default: 2). This makes the execution a lot faster on multi-core machines.

Run analysis

Start via batch_start.R. Analysis profits from multi-core machines. To run this from the command-line (and not interactively from Rstudio), do R --no-save <src/batch-start.R 2>&1 |tee out.log