Skip to content

cucu-o0/titanic

Repository files navigation

Titanic

Machine learning from disaster

This is the legendary Titanic ML kaggle competition

titanic

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.

About

Machine Learning from Disaster

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published