Winter Week 4 Blog Post Chongxian Chen

This week Jake collected a lot of data for NCAA matches. And after learning from classification examples I was able to read all the csv data file successfully into numpy array. I also tried using sklearn to train the model and produce a test outcome. The outcome basically looks fitting to the data, i.e the predicted outcome relates to the expected outcome mostly correct. But the prediction also generate some extreme confidence like 99% of winning chance. Which I think could be problematic since extreme prediction will have a extreme penalty in LogLoss equation. We will need to be more careful with extreme prediction. Next week I will look into the extreme prediction in more detail to make our predictions more reasonable.