scientific article; zbMATH DE number 1069600

From MaRDI portal
Publication:4356562





We investigate a class of parametric models, the integer-valued autoregressive schemes, which makes allowance for the discreteness of the data while retaining some of the properties of the conventional autoregressive models, so-called INAR models. We prove some asymptotic results for the conditional maximum likelihood parameter estimates of INAR(1) processes. Then, we illustrate the usefulness of the model for a special case and apply the estimation procedure to some data coming from an extensive study concerned with the effectiveness of a certain antiepileptic drug. Here, a large number of patients recorded for approximately half the year the daily numbers of epileptic seizures, first with only standard medication being applied, then with additional use of the new drug in the test group and a placebo in the control group. Comparison of the total number of fits before and after treatment and between test and control group showed the general effectiveness of the drug against particular kinds of seizures. The time-ordering of the data was, however, not used in the analysis, and the question was, if it contains more information about how the reduction of the number of fits is achieved. For the entire collection see [Zbl 0846.00010].




Cited in
(40)








This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4356562)