Files of my Titanic Kaggle Competition.
- Titanic datatreatment: prepares the data to be fed to the models. Dummies categorical, scales numerical, engineers new features.
- Titanic EDA: Exploratory Data Analysis of the treated data.
- Titanic CV model selection: a comprehensive grid parameter scan of different models. It outputs data-frames (into CSV files) of the results for different parameters.
- DNN CV model selection: grid parameter scans for multiperceptron models, using Keras and SciKit.
- CV analyser: a notebook to study the results of Titanic CV model selection.
- Titanic Submission Predictions: The generator of prediction files. It employs the best estimators assessed at CV analyser, and produces further ensemble predictions with voting.
- Titanic Deck Predictor: an attempt to predict the deck to fill missing values. These turn out to worsen the predictions for survival chances and are therefore ignored for now.