. Predicting passengers who survived the Titanic disaster using a decision tree.

Abstract

The article proposes the use of the decision tree method for predicting the passengers who survived the Titanic liner disaster. The Titanic - Machine Learning from Disaster dataset, which is publicly available on the Kaggle platform, is used as input. Decision trees are well suited for solving classification and forecasting problems, and their ease of interpretation makes this method the best choice among other machine learning algorithms. On training data set, modification has been performed to fill the missing values. The distribution of qualitative and quantitative data features and the search for patterns in the data were evaluated using visual data analysis, which allowed us to identify the passenger features that correlate with their survival the most and improve the data set accordingly. The decision tree for the final dataset was built using the scikit-learn library (sklearn), which provides powerful tools for machine learning Python. The accuracy of the built decision tree is 77% of the deferred sample. Further study of the application of the decision tree for this dataset can be done by using the hyperparameter tuning method, which will help to improve the accuracy of the constructed decision tree.