We develop a general procedure to fit a splicing model to censored and/or truncated data. We propose to combine the flexibility of the mixed Erlang distribution to model the body of the distribution with the Pareto distribution to provide a suitable fit for the tail.
We present a flexible and effective fitting procedure for multivariate mixtures of Erlangs, that is able to deal with randomly censored and fixed truncated data.
We present a calibration procedure for fitting mixtures of Erlangs to censored and truncated data by iteratively using the EM algorithm.
We demonstrate our implemented fitting procedure and graphical tools built in R for univariate and multivariate mixtures of Erlangs in several applications in the context of loss modelling.