Group members: Siyao Chang (changsiyao
),
Boying Gong (boyinggong
),
Benjamin Hsieh (BenjaminHsieh
),
Brian Qiu (brianqiu
),
Jiang Zhu (pigriver123
)
Topic: [The Neural Basis of Loss Aversion in Decision Making Under Risk] (https://openfmri.org/dataset/ds000005/)
The original Science paper found
here was written by
Sabrina M. Tom, Craig R. Fox, Christopher Trepel, and Russell A. Poldrack at
UCLA. The main aim is to identify regions of brain related to loss aversion in
decision making (in this case a gambling task of 50/50 chance of gaining/lossing
money) and investigate the behavorial and neural correlates of loss aversion.
In our project we aim to recreate a subset of the paper results with the goal
of of checking model assumptions, creating analysis pipelines of fMRI research,
and comparing results with Tom's (et al) paper. Please see paper report in
project-theta/paper
for further discussion.
To generate paper please navigate to paper
directory and make all
.
Follow the instructions below to reproduce/update analyses figures.
Note: These steps, in particular make all
, can take dozens of hours due
to the size and nature of the analyses. We do not recommend running it but
instead follow the instructions below Alternate Steps for specific desired
results.
make resultfolder
will generate the results folderproject-theta/results
to save all analysis output and figures generated from running scripts. Please note that this folder will be git ignored (exist on local repository). If one decides tomake all
or run any analyses inscripts
,
run this command beforehand.make all
will generate everything related to the analyses. It will pull in and validate data inside the data directory, create images saved inside the results directory, move relevant images to the paper/figures subdirectory, and generate the final report report.pdf inside the paper directory. The process can take a VERY long time as the data is over 16 GB and the analyses take quite a while as well. To save time one can view the final report without running this command by going to the paper directory and running make all from there since the needed figures are already preloaded there.
- First,
make data
to download, unzip, and validate data, saving datasets to the data directory. Note: Run time around 30 minutes. - Second,
make resultfolder
to initialize analyses output directory. - Navigate to
code/scripts
and run the specific desired analysis. - Navigate to
code
directory andmake paperfig
to update paper figures - Navigate to
paper
directory andmake all
to generate paper.
make clean
will delete extraneous files created by our scriptsmake coverage
will check the coverage of our data, code directoriesmake test
will run the tests in ourcode/utils/tests
andcode/data/tests
directorymake verbose
is the verbose version of 'make test'
code
contains the functions and scripts for analysis and graphingdata
holds the relevant datasets that are pulled in and validated either through make commands there or make all here.paper
contains the final report, report.pdf, once the appropriate make commands are run, either here or in paper. It also has the relevant figures for the paper.slides
contain our presentation slides used in class (Stat 159/259).requirements.txt
is the a text file listing the required package versions.tools
contains the material related to travis ci.
Many thanks to Jarrod Millman, Matthew Brett, J-B Poline, and Ross Barnowski for their advice and efforts that made this project possible.