Statistical Learning

Comparison of change point detection methods

This post compares a few change point detection method available in R given different time series dynamics and research questions. Change points or breakpoints are abrupt variations in time series data and may represent transitions between different states. The detection of change points is useful in modelling and prediction of time series and is found in application areas such as medical condition monitoring, speech and image analysis or climate change detection.

Continue reading

How to normalize the RMSE

This post has been stimulated by a discussion with a colleague who asked about the normalization method for the root mean square error (NRMSE) in the INDperform R package, which is based on the indicator testing framework outlined in my article (Otto et al. 2018)1. At the time of writing the article and package I simply used a common approach and didn’t test it much further. But sparked by this discussion I started to test it thoroughly (as you will see below), which will make me revise the package.

Continue reading