Titanic Survival Predictor

Machine learning can sound very mysterious at times, and it is not always clear what you can do with it. Therefore, I created the Titanic Survival Predictor to show the power of data science, and more precisely, machine learning. 

In this project, I used a Kaggle dataset that is used by many data scientists to improve their machine learning skills. The goal of this dataset is to predict whether a person survived the titanic based on passenger information, such as age, passenger class, embarking port and number of relatives on board. The final product is a Heroku-app to find out if you would have survived the Titanic disaster. 

As with most datasets, the data first needed to be cleaned, and a selection of information needed to be made, as garbage in = garbage out. I then enriched the data through feature engineering, and scaled and encoded the data for the data preparation. Since there is no “1 model fits” all in machine learning, I tried out 5 different classification models and scored them all on the accuracy of predicting whether the passenger would have survived the Titanic disaster. The best scoring model was a Random Forest with an accuracy of 78% on the Kaggle test dataset. To allow others to interact with the Titanic prediction model, I created a Heroku-app that predicts the survival probability based on user inputs. 

Would you have survived the titanic disaster?
Find out here: https://too-titanic-app.herokuapp.com/