A function that given a fitted singleR class object computed marginal frequencies by as sum of probability density functions for each unit in data at each point i.e. kth element of marginal frequency table is given by \mjseqn \sum _j=1^N_obs\mathbb P(Y_j=k|\eta _j). For k=0 only (if specified at call) they are computed as \mjseqn \hat N-N_obs because \mjseqn \boldsymbol f_0 is assumed to the unobserved part of the studied population.
These frequencies are useful in diagnostics for count data regression, such as assessment of fit.