This paper proposes a new minimum description length procedure to detectmultiple changepoints in time series data when some times are a priori thoughtmore likely to be changepoints. This scenario arises with temperature timeseries homogenization pursuits, our focus here. Our Bayesian procedureconstructs a natural prior distribution for the situation, and is shown toestimate the changepoint locations consistently, with an optimal convergencerate. Our methods substantially improve changepoint detection power when priorinformation is available. The methods are also tailored to bivariate data,allowing changes to occur in one or both component series.
展开▼