Titanic
Tabular Data
Data Preparation
Classification
Neural Nets
Hyperparameter optimization
Competition
fastai
XGBoost
Predict who survived the Titanic shipwreck in a kaggle competition.
| LName | Pclass | Sex | SibSp | Parch | Cabin | Title | Age | Fare | target |
|---|---|---|---|---|---|---|---|---|---|
| Vovk | 3 | male | 0 | 0 | N | Mr | -0.5500 | -0.4892 | 0 |
| Tobin | 3 | male | 0 | 0 | F | Mr | -0.0827 | -0.4921 | 0 |
| Porter | 1 | male | 0 | 0 | C | Mr | 1.2989 | 0.3984 | 0 |
| Wick | 1 | female | 0 | 2 | C | Miss | 0.1156 | 2.6696 | 1 |
| Sutehall | 3 | male | 0 | 0 | N | Mr | -0.3281 | -0.5062 | 0 |
| Klaber | 1 | male | 0 | 0 | C | Mr | 0.8885 | -0.1138 | 0 |
This dataset contains information about all passangers of the Titanic voyage. After preprocessing and condensing some features I used fastai to train a Neural Net on that data. To find the “best” hyperparameters I ran a random search and stuck with the best performing. I also trained a Gradient Booster to compare the results.