Time spent: 5h
Total: 31.5h/10000h
Today I did my first Kaggle competition, (obviously) the Titanic dataset. I got a 0.68 accuracy score and after looking up some feature engineering techniques I managed to lift it up to 0.75. I also went deeper into SVMs to see how they really worked, and tried to develop more geometric intuition for them.
Overall I’d argue this was my most productive day of ML thus far. I’ll probably end up writing a post about SVMs soon.