Notes on the Kaggle Titanic Stacking Model

Thu, 17 Aug 2017 11:51:08 +0800

While reading through Kaggle kernels for the Titanic challenge, many of them use SVM, RandomForest, LogisticRegression, etc. What makes this particular kernel interesting is that it builds a model from six different learners:
Introduction to Ensembling/Stacking in Python Using data from Titanic: Machine Learning from Disaster
At Level 1 it uses RandomForestClassifier, AdaBoostClassifier, GradientBoostingClassifier, ExtraTreesClassifier, and SVM, and at Level 2 it uses XGBoost. I sketched the overall flow of the model to make it easier to understand — the raw source is hard to parse quickly. The author cleverly uses classes to keep the notebook code clean, which also makes it easier to modify and organize later.

Stacking - Tag - 300.Watts

Notes on the Kaggle Titanic Stacking Model