Yansen Sheng

Santander Customre Satisfaction Predictor

Introduction:

• This is a gradient decision tree trained by me and two other teammates, Yizhou Feng and Shuangchen Shen, for a Kaggle competition.
• The link to the competition is here: https://www.kaggle.com/c/santander-customer-satisfaction
• This gradient decision tree uses extreme gradient boosting model (http://xgboost.readthedocs.io/en/latest/model.html).
• Irrelevant features (such as zero values and uniform values) are ignored during the training phase to reduce the amount of calculation.
• Features that are not quite relevant are also excluded by using an Extra Trees Classifier.
• Link to this project is here