Simple ShinyApp in R for the February 2015 SETT article about machine learning.
This application provides some knobs to adjust and explore goodness of fit for a logistic regression model. Some random data is provided and a logistic regression model trained with the parameters set by the user. A confusion matrix and derived measures are displayed as the inputs are changed. The ability to see the confusion matrix and performance for both the validation data (default) and training data is provided so the user can observe the difference between the trained model used on the data it was trained with as well as the previously unseen validation data.
The plot includes the two level classification data sets and a plot of the decision contour for the model. The plot is regenerated by user command to reduce the amount of time spent regenerating the decision contour. This contour is made by creating a grid (default 100x100) of points in the parameter space and making a prediction at each point. A contour is then made to show where the decision for classification will change from one value to the other.