University of Copenhagen

The R-package ppstat implements methods for point process statistics for multivariate point processes (that is, marked point processes) on the line. The package is still under development. Version 0.8 is available from CRAN. The development version is available from R-Forge.

The package is based on another R-package processdata. This package implements a data structure for storing point process and general stochastic process data. It includes various plotting and subsetting facilities.

The ppstat package implements a framework termed generalized linear point process models where the model is specified in terms of a (non-linear) transformation of a linear combination of predictor processes. The package supports a specification of such models via a formula, which can be used to express a number of different transformations and filters of the basic processes that enter into the predictor. The data structure supports observations of one or several independent point processes, whose intensity can be given in terms of its own internal history as well as additional covariate processes.

The package was originally developed for point process modeling of genome features. However, there is nothing in the package that is specific to genomic modeling. Indeed, that package implements models that have been used to model such diverse things as financial trade times and neuron spike times.

License: GPLv3

Comments and bug reports

Other programs

  • To install the package from CRAN run
    from the command line. Make sure that dependencies get installed.

  • To install the development version running the most recent version of R (discouraged for production usage) run
    install.packages("processdata", repos="") 
    install.packages("ppstat", repos="") 
    You will most likely have to install other dependencies yourself.

    To use the package you will have to construct data objects of class MarkedPointProcess containing your data. The demo examples shows some examples. To see the results of running the demo type
    and to get the location of the examples.R file type
    system.file("demo", "examples.R", package="processdata")
    See also
    for information on the construction of marked point processes. Demos
    from ppstat show examples of the specification and estimation of generalized linear models.

NOTE: These older versions are not recommended for usage. They use a different implementation of the data structures, and many central methods have been altered or renamed. These implementations were used for the paper Multivariate Hawkes process models of the occurrence of regulatory elements and are kept here for the record.