A solution to the Titanic ML competition.
Steps
- First we acquire the data and check them in order to understand them and see if there are any null/missing values
- Then we analyze the data, first by pivoting and then with visualization, in order to understang which will help us get a better solution and how they are connected.
- Now we correct the data by dropping features we don't need, creating new from existing, converting categorial, completing numnerical features and creating new by combining existing ones.
- After finnishing with the data we create various models and test them.
- Finally we evaluate them to find the one with the best score.
To solve this problem I did my own research trying various combinations and preprocessing of the data. Finally I used some technics that Manav Sehgal used in his solution to achieve even better scores.