Last week I talked about using a random classifier model for continuous out-of-sample predictions using the past three months of OS performance data to make a decision about whether a system is or is not likely to be profitable for the following three months. With the positive results we obtained with this approach in validation sets […]