pjankiewicz / asap-sas Goto Github PK

Code for 5th place in ASAP-SAS competition

License GPLv3: http://www.gnu.org/copyleft/gpl.html Authors: Jonathan Peters, Paweł Jankiewicz

1. Installation.
   • Python 2.6.6+ Libraries needed: numpy, nltk*, pyaspell
   • R 2.14.1+ Libraries needed (as in config.r file)
   • RStudio 0.95.262+

• for nltk there is a necessity to download some external data: nltk.corpus.stopwords nltk.corpus.brown nltk.corpus.wordnet

  It is done very quickly using a command nltk.download() and selecting those 3 libraries

2. Place the private leaderboard data set as private_leaderboard.tsv file in data/raw/ folder The file that resides there is only a dummy file that helped us create a process.

3. Change the header to setwd(YOUR_CURRENT_PATH) in files:

   • 1c_prepare_features.r
   • 2_prepare_data.r
   • 4_make_models.r
   • 5_blend.r

4. Run the files:

   • 1a_prepare_essays.py (python)
   • 1b_prepare_ngrams.py (python)
   • 1c_prepare_features.r (Rstudio)
   • 1d_prepare_chunks.py (python)
   • 1d_prepare_essay_specific_features.py (python)
   • 1g_prepare_features.py (python)
   • 2_prepare_data.r (Rstudio)
   • 3_make_models.r (Rstudio)
   • 4_blend.r (Rstudio)

It will take 1-2 weeks to calculate all the models depending on how many cores you have.

5. The submission should be placed in submissions/private.csv