Ensemble Models

This section describes experiments using ensemble models that combine the probability outputs from logistic regression, random forests, and neural net models, and use those either as the sole inputs for a new model, or alongside the normal features in the training and test set.

This section contains the following notebooks (all experiments are performed with a single training/test split):

• Ensemble model fitting - generate probabilities: Generates the prediction probabilities from logistic regression, random forests, and a neural network.

• Comparison of model outputs Compares the outputs of three types of model.

• Ensemble combine with logistic regression: combine output from logistic regression, random forests, and neural network models in a single logistic regression model, with or without original data features.

• Ensemble combine with random forests: combine output from logistic regression, random forests, and neural network models in a single random forests model, with or without original data features.

• Ensemble combine with neural network: combine output from logistic regression, random forests, and neural network models in a single neural network model, with or without original data features.

• Summary of results from ensemble models: summary of ensemble model accuracies.