Titanic: Machine Learning from Disaster

This my entry for the Titanic competition on Kaggle. May 2019: public score is 0.80382, which is a top 10% ranking on the leader board of around 11.249 participants.

I wrote a Python script to automate the following tasks: read data from CSV, perform data wrangling, give a brief data analysis, impute missing values, add predictive value with feature engineering, train stacked ensemble (build with h2o framework) and generate predictions on unseen test data. Click on the link to read the Python code: a-quick-dive-into-h2o-with-python.py.