Skip to content

markschwarz/amazonaccess

 
 

Repository files navigation

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.

Usage:

[python] classifier.py [-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 classifier.py -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] logistic.py log 75
[python] ensemble.py

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.

Requirements:

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.

License:

This content is released under the MIT Licence.

About

Amazon Employee Access Challenge

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published