Need help with amazonaccess?
Click the “chat” button below for chat support from the developer who created it, or find similar developers for support.

About the developer

201 Stars 136 Forks MIT License 22 Commits 1 Opened issues


Amazon Employee Access Challenge

Services available


Need anything else?

Contributors list

# 227,220
22 commits

Amazon Employee Access Challenge

This code was written by Paul Duan ([email protected]) and Benjamin Solecki ([email protected]). It provides our winning solution to the Amazon Employee Access Challenge. Our code is currently not merged. You'll find Benjamin's code in the BSMan/ folder, which needs to be run separately.


[python] [-h] [-d] [-i ITER] [-f OUTPUTFILE] [-g] [-m] [-n] [-s] [-v] [-w]

Parameters for the script.

optional arguments: -h, --help show this help message and exit -d, --diagnostics Compute diagnostics. -i ITER, --iter ITER Number of iterations for averaging. -f OUTPUTFILE, --outputfile OUTPUTFILE Name of the file where predictions are saved. -g, --grid-search Use grid search to find best parameters. -m, --model-selection Use model selection. -n, --no-cache Use cache. -s, --stack Use stacking. -v, --verbose Show computation steps. -w, --fwls Use metafeatures.

To directly generate predictions on the test set without computing CV metrics, simply run:

python -i0 -f[output_filename]

This script will launch Paul's model, which incorporates some of Benjamin's features. Benjamin's model is in the BSMan folder and can be run this way:

(in BSMan/)
[python] log 75

The output of our models is then combined by simple standardization then weighted averaging, using 2/3 Paul's model and 1/3 Benjamin's.


This code requires Python, numpy/scipy, scikit-learn, and pandas for some of the external code (this dependency will be removed in the future).
It has been tested under Mac OS X with Python v.7.x, scikit-learn 0.13, numpy 0.17, and pandas 0.11.


This content is released under the MIT Licence.

We use cookies. If you continue to browse the site, you agree to the use of cookies. For more information on our use of cookies please see our Privacy Policy.