***
I continued to work on some teachable methods for refining a machine learning project today. Once I got myself organized, I found it a much more productive and enjoyable experience. I still haven't done a second submission on my data science competition, but I have developed a way of organizing data science projects, figured out how to put together a scikit-learn pipeline, and how to use GridSearchCV to find optimal hyperparameters for a machine learning algorithm. I also figured out how to implement a custom KFold cross validation scheme and how to get it to use a log loss scoring system. Needless to say, other than my run/walk this morning, I didn't really get out much today. I feel a bit slothful (at least physically).
No comments:
Post a Comment