Starter notebook is in progress.

[wakame1367]

What content should be included? below link is draft. https://drive.google.com/open?id=1aSiplVhB9Hjcj8Ib-A9zkAjEgykpllH4

[nyanp]

Oh, cool! Since nyaggle is a collection of independent utility functions, I think we don't need to visit all modules (this use case should rather be covered by the API reference).

If the notebook covers an end-to-end modeling process in typical tabular data competition, it would be very valuable for data-science beginners. In that case, for example, the content would be consist of the following steps:

1. load train.csv, test.csv, sample_submission.csv
2. manual feature engineering
3. save/load features using feature_store (cached_feature decorator is suitable for a single notebook)
4. (optional) check trian/test difference by adversarial_validation
5. call run_experiment with mlflow
6. see the result in the notebook and localhost:5000

The following notebook will be helpful as examples with similar purpose.

• featuretools with Kaggle HomeCredit competition
• H2O.ai
• PyTorch

[wakame1367]

Starter notebook moved. nyaggle-tutorial-for-titanic.ipynb