Constrained changepoint GUI

A Shiny web application for the gfpop R package. [...]

Detecting sudden changes in data is important to a variety of fields. For example, cardiologists may look for sudden changes in electrocardiogram traces to diagnose heart problems and financial analysts may look for sudden changes in certain stock prices to predict market shifts. While these so-called ‘changepoints’ are generally apparent by eye, they remain challenging to predict computationally.

Beginning in 2017, Dr. Toby Hocking and others presented a log-linear algorithm called General Functional Pruning Optimal Partitioning (GFPOP) that can efficiently and accurately detect changepoints. They also demonstrated how GFPOP can be used for a variety of practical applications, such as classifying human tumors and learning where proteins interact with DNA.

To make GFPOP more accessible to the researchers that may benefit from the algorithm, Hocking and others created the gfpop R package. During the summer, I am building a graphical user interface (GUI) for the gfpop R package to further increase its utility and accessibility to a general audience.