Deployed Soldiers Classification Experiment

February 20, 2015
This experiment uses a simple dataset to demonstrate the ability to predict soldiers resigning after their deployment
Deployed Soldiers Classification Experiment\n===========================================\nThis experiment uses a simple fictitious dataset that represents US Soldiers who are deployed downrange and attempts to predict whether they will leave the Army after their deployment (represented in the training dataset by a 1 in the \"**Active**\" column.\n\n\nData Flow\n---------\nThe experiment starts with a **Convert to Dataset** object (which is not entirely necessary for this dataset, but it's good practice to include it and remove missing values) and then uses the **Project Columns** object to remove unnecessary columns. The experiment then uses 2 split objects to create 3 datasets of 60%, 20% and 20% for Training, Validation and Testing. The experiment uses 2 models to train and then the **Sweep Parameters** object to find the best combination of parameters. Finally the models are scored and evaluated to determine which model performs best.