Spaceship Titanic Competition

This is a repository for the Spaceship Titanic Competition in Kaggle, used by the Kaggle 1 Insper Dynamics team. This repository will be mostly used for practicing and testing the code for future, more complex competitions.

Members

• Felipe Maluli de Carvalho Dias
• Lucca Hiratsuca Costa
• João Alfredo Cardoso Lamy
• Thomas Chiari Ciocchetti de Souza

Submissions

• Submission1: Uses a Random Forest Classifier, with 100 trees, using scikit-learn library (more info on the notebook Spaceship-Titanic.ipynb).
• Submission2: Uses a Logistic Regression Classifier, with scikit-learn library, and applies Feature Selection and Engineering techniques (more info on the notebook Spaceship-Titanic-LogReg.ipynb).
• Submission3: Uses a Random Forest Classifier, using tensorflow-decision-forests library, with default parameters and configurations (more info on the notebook ST_TFDF.ipynb).

Images

• tree_1.png and tree_model_1.png: Represent the Decision Tree Classifier first elaborated in the notebook Spaceship-Titanic.ipynb, with max_depth=3.
• tree_model_2.png: Represents the Decision Tree Classifier elaborated in the notebook Spaceship-Titanic-LogReg.ipynb, with max_depth=5.
• tree_random_forest.png: Represents one of the trees of the Random Forest Classifier elaborated in the notebook Spaceship-Titanic.ipynb, with n_estimators=100.