top secret

###Setup Create .env file in the project root folder. It needs to look like this:

KAGGLE_USER='my kaggle login'
KAGGLE_PASSWORD='my kaggle password'

Put this directory on your python path like this:

export PYTHONPATH="${PYTHONPATH}:PATH_TO_THIS_DIRECTORY"

Put it in your bash_profile (or equivalent) to make it permanent.

then run

weremeerkat/data/download_data.sh

To get the good stuff. Code for testing the models and making submissions lives in src/models/models and src/models/benchmarks at the moment. The idea is to define all the reusable components there and to do one-off experiments using those components in src/models/run.py.

python weremeerkat/models/run.py

To put the csv filer in a sqlite database run:

weremeerkat/data/create_database.sh

This will create data/interim/database.db and import all the tables into it. To instead save everything as spark friendly parquet files do:

spark-submit weremeerkat/data/make_parquet.py

You must have spark installed and available on path. Then make train/test split by running

spark-submit --driver-memory 25g weremeerkat/data/train_test_split.py

Next to create training set for libffm, run:

spark-submit --driver-memory 25g weremeerkat/data/features/build_features.py

This will create a bunch of files in data/interim/features/basic - full train set, test set, and train set split 80-20 for cross validation. The idea is that each new set of features new_features will create files in data/interim/features/new_features and the downstream processing can be the same.

Next run libffm:

python weremeerkat/models/run_benchmark.py basic

This will train the model and save it in models/basic and make predictions in data/processed/basic. The basic argument refers to the training set created with the previous command. Next, evaluate the results and make submission file by running:

python weremeerkat/models/run_benchmark.py basic

Submission file and a file with evaluation metrics will appear in data/processed/basic

Project Organization

├── LICENSE
├── Makefile           <- Makefile with commands like `make data` or `make train`
├── README.md          <- The top-level README for developers using this project.
├── data
│   ├── external       <- Data from third party sources.
│   ├── interim        <- Intermediate data that has been transformed.
│   ├── processed      <- The final, canonical data sets for modeling.
│   └── raw            <- The original, immutable data dump.
│
├── docs               <- A default Sphinx project; see sphinx-doc.org for details
│
├── models             <- Trained and serialized models, model predictions, or model summaries
│
├── notebooks          <- Jupyter notebooks. Naming convention is a number (for ordering),
│                         the creator's initials, and a short `-` delimited description, e.g.
│                         `1.0-jqp-initial-data-exploration`.
│
├── references         <- Data dictionaries, manuals, and all other explanatory materials.
│
├── reports            <- Generated analysis as HTML, PDF, LaTeX, etc.
│   └── figures        <- Generated graphics and figures to be used in reporting
│
├── requirements.txt   <- The requirements file for reproducing the analysis environment, e.g.
│                         generated with `pip freeze > requirements.txt`
│
├── src                <- Source code for use in this project.
│   ├── __init__.py    <- Makes src a Python module
│   │
│   ├── data           <- Scripts to download or generate data
│   │   └── make_dataset.py
│   │
│   ├── features       <- Scripts to turn raw data into features for modeling
│   │   └── build_features.py
│   │
│   ├── models         <- Scripts to train models and then use trained models to make
│   │   │                 predictions
│   │   ├── predict_model.py
│   │   └── train_model.py
│   │
│   └── visualization  <- Scripts to create exploratory and results oriented visualizations
│       └── visualize.py
│
└── tox.ini            <- tox file with settings for running tox; see tox.testrun.org

---

Project based on the cookiecutter data science project template. #cookiecutterdatascience