This is going to grow big.
- Titanic: Predict survival on the Titanic and get familiar with ML basics (Titanic: Machine Learning from Disaster)
- Nuclei: Find the nuclei in divergent images to advance medical discovery (2018 Data Science Bowl)
- use
pipenv
in project dirs - for
keras
andh5py
make sure you have libhdf5 installed:sudo apt-get install libhdf5-dev
- create new competition folder structure by
create_new_competition.sh
. - use kaggle API to manage data and submissions:
cd titanic kaggle competitions list -s titanic # get info about competition kaggle competitions files -c titanic # list data files kaggle competitions download -p data -c titanic # download files kaggle competitions submit -f results/result.csv -m "My message" -c titanic
- export jupyter notebooks to python script:
jupyter nbconvert --to script [YOUR_NOTEBOOK].ipynb
- run with
ipython [YOUR_NOTEBOOK].py
(python doesn't support jupyter cell magic)
- run jupyter notebook from command line:
jupyterrun [YOUR_NOTEBOOK].ipynb # custom function: creates temporary .py script and executes it with ipython