Fractional factorial experiments often produce ambiguous results due to confounding among the factors; as a consequence more than one model is consistent with the data. Thus, the practical problem is how to choose additional runs in order to discriminate among the rival models and to identify the active factors. The R package OBsMD solves this problem by implementing the objective Bayesian methodology proposed by Consonni and Deldossi (2016). The main feature of this approach is that the follow-up designs are obtained through the use of just two functions, OBsProb() and OMD() without requiring any prior specifications, being fully automatic. Thus OBsMD provides a simple tool for conducting a design of experiments to solve real world problems.
- Bayesian design of experiments
- Bayesian model selection
- Model discrimination
- Screening experiments