Kaggle : Titanic - Machine Learning from Disaster

Summary

Titanic ML competition from Kaggle

More info : visit https://www.kaggle.com/c/titanic

Purpose : Predict which passengers survived

Titanic shipwreck

More info about Titanic shipwreck : visit https://en.wikipedia.org/wiki/Sinking_of_the_Titanic

Repository content

• titanic.ipynb : Jupyter notebook
• train.csv: training dataset from Titanic-Machine Learning from Disaster https://www.kaggle.com/c/titanic
• test.csv: test dataset from Titanic-Machine Learning from Disaster https://www.kaggle.com/c/titanic
• my_submission.csv : predictions for test dataset

Titanic Kaggle competition

Notebook

• Data visualisation
• Pre-processing with pipeline
  • Imputer
  • One-hot encoding

• Machine learning
  • Model : Random Forest Classifier
  • Evaluation of the accuracy score
  • Optimisation of model parameters with Grid Search CV

Competition predictions

Score = 0.79425