Kaggle Titanic competition log (updated from time to time)

2018/9/15
signed up for the Titanic competition. The competition is marked as Knowledge, and it is purely for entry purposes and has no bonuses.
After downloading the data, I tried to fit the most basic DTR, but failed. The reason is that the feature'Age' has a missing value, which is not acceptable to DTR.
According to the instructions on the competition page, I found a speedml package and the corresponding solution.
But before that, I first refer to https://blog.csdn.net/u013115001/article/details/78012553 for guidance and do the introductory exercises.
The first step is to deal with missing values ​​and character values. Refer to pandas introduction and anonymous function lambda .
The second step is to change DTR to DTC to obtain discrete values. And changed median() to mean(), and got a better score of 0.73684. Different from the original score, it is speculated that the data may have been updated.