No description, website, or topics provided.
Clone or download
Latest commit eda48cb Feb 25, 2015
Type Name Latest commit message Commit time
Failed to load latest commit information.
bag initial commit Feb 25, 2015
base initial commit Feb 25, 2015
ensemble initial commit Feb 25, 2015
README initial commit Feb 25, 2015 initial commit Feb 25, 2015
license.txt initial commit Feb 25, 2015 initial commit Feb 25, 2015 initial commit Feb 25, 2015
tr.rx.csv initial commit Feb 25, 2015
va.rx.csv initial commit Feb 25, 2015


4 Idiots' Approach for Click-through Rate Prediction

Our team consists of:
    Name              Kaggle ID         Affiliation
    Yu-Chin Juan      guestwalk         National Taiwan University (NTU)
    Wei-Sheng Chin    mandora           National Taiwan University (NTU)
    Yong Zhuang       yolicat           National Taiwan University (NTU)
    Michael Jahrer    Michael Jahrer    Opera Solutions

Our final model is an ensemble of NTU's model and Michael's model. Michael's
model is based on his work in Opera Solutions, so he cannot release his part.
Therefore, in the codes and documents we only present NTU's model.

This README introduces how to run our code up. For the introduction to our
approach, please see

The model we use for this competition is called `field-aware factorization
machines.' We have released a package for this model at:

System Requirement

- 64-bit Unix-like operating system

- Python 3

- g++ (with C++11 and OpenMP support)

- pandas (required if you want to run the `bag' part. See `Step-by-step'


Our solution is an ensemble of 20 models. It is organized into the following
three parts:
    name       public score     private score     description        
    base             0.3832            0.3813     2 basic models

    bag              0.3826            0.3807     2 models using bag features.

    ensemble         0.3817            0.3797     an ensemble of the above 4 
                                                  models and 16 new small models

Because the `bag' part consumes a huge amount of memory (more than 64GB), and
the `ensemble' part takes a long time to run, this instruction guides you to
run our `base' part first. If you want reproduce our best result, please run the
commands in the final step on a suitable machine.

1.  First, please use the following command to run a tiny example up

    $ ./ x

2.  Create a symbolic link to the training dataset

    $ ln -sf <training_set_path> tr.r0.csv

3.  Add a dummy label to the test set

    $ ./ <test_set_path> va.r0.csv

4.  Checksum

    $ md5sum tr.r0.csv va.r0.csv
    f5d49ff28f41dc993b9ecb2372abb033  tr.r0.csv
    6edd380a5897bc16b61c5a626062f7b3  va.r0.csv

5.  Reproduce our base submission

    $ ./ 0
    Note: base.r0.prd is the submission file

6.  (optional) Reproduce our best submission

    $ ./ x

    If success, then run

    $ ./ 0

    Note: The algorithm in the `bag' part is non-deterministic. That is, the
    result can be slightly different when you run it two or more times.


If you want to trace these codes, please be prepared that it will take some
efforts. We do not have enough time to polish the codes here to improve the
readability. Sorry about it. 

For any questions and comments, please send your email to:

    Yu-Chin (