Main Experiments (Sections 6.2 and 6.3)

In the following, <dias-benchmarks root> is the root of dias-benchmarks. For example, if you extract 10.1145_3639313_source_code.zip in the directory /home/foo, you should get the directory /home/foo/artifact. Then, <dias-benchmarks root> is: /home/foo/artifact/dias-benchmarks.

To produce all the results needed you only need to run two scripts. First, cd <dias-benchmarks root> and run the following. It's necessary to source the script!:

source setup_everything.sh

This should set up everything needed to run the experiments, but in case it fails, you can consult <dias-benchmarks root>/SETUP.md.

To run the experiments in the paper, cd <dias-benchmarks root>/runner and run:

./paper_exps.sh

Producing the plots

Running paper_exps.sh should produce the following directories:

stats-rewr_OFF-modin_12-repl_LESS-sliced_exec_ON
stats-rewr_OFF-modin_4-repl_LESS-sliced_exec_ON
stats-rewr_OFF-modin_8-repl_LESS-sliced_exec_ON
stats-rewr_OFF-modin_OFF-repl_LESS-sliced_exec_ON
stats-rewr_OFF-modin_OFF-repl_STD-sliced_exec_ON
stats-rewr_ON-modin_OFF-repl_LESS-sliced_exec_ON
stats-rewr_ON-modin_OFF-repl_STD-sliced_exec_OFF
stats-rewr_ON-modin_OFF-repl_STD-sliced_exec_ON
stats-rewr_stats

We move those to <dias-benchmarks root>/stats (where stats.ipynb is) e.g., with:

cd <dias-benchmarks root>
mv runner/stats* stats

Then, you can just run stats.ipynb to produce the plots.

Comparing Various Dataframe Libraries (Section 6.6)

The directory <dias-benchmarks root>/pandas-replacements is independent and ha the code used to produce the numbers in Table 2 of the paper. Producing the numbers and the table was done manually (FYI, the table was created with Google Docs).