Predicting Titanic Survivors by Using Machine Learning

Abstract

About a century ago, one memorable night in April 1912, a world-shattering event happened. The Titanic, the 2,240-passenger luxury cruise ship, sank forever off the coast of Newfoundland in the North Atlantic after extensive damage to its hull by an iceberg on its maiden voyage. Only 705 people survived this disaster. Although nearly a century has passed, the research on Titanic has never stopped, and there are still many studies on it. This study was supposed to predict the survival of passengers on Titanic using different methods based on data from the Kaggle competition "Titanic: Machine Learning from Disaster." It predicted each passenger in the test set who would survive the sinking. The result was the percentage of correct prediction. In the Machine Learning study, the task is to achieve 80% accuracy in predicting the survival distribution of the Titanic disaster based on the demographic data testing notebook by different algorithms models. Using classification is the main point to calculate the efficiency achieved by those models through the test environment. The f-measurement scores obtained from the machine learning technology were in comparison with the f-measurement scores obtained by Kaggle.