Titanic

Machine learning from disaster

This is the legendary Titanic ML kaggle competition

Goal

Create a model that predicts the probability of survival of passengers in a shipwreck.

Model selection

Selection of classification algorithms:

1. Logistic Regression
2. Gaussian Naive Bayes
3. K-nearest Neighbors
4. Linear Support Vector Machine
5. Random Forest

The Winner

Random Forest > Multiple Random Decision Trees

Conclusions

Chances to survive:

~80% accuracy
~71% f1-score
~82% ROC-AUC score

The model is quite good in predicting the probability of survival of the passengers but there is still room for improvement.