In many cohort studies, we may observe repeated outcomes with competing events. Events may occur within an interval or time to events itself are discrete. The occurrences of events at different stages produce a trajectory of events for study subjects. Objectives are to model such data efficiently and prediction of unconditional trajectory probabilities for a subject with specified covariate vectors. We extended the binary regressive logistic model for multinomial outcomes and proposed a framework to predict unconditional trajectory probabilities. Also, multistate Markov model is used for the same and results are compared. The regressive modelling approach is very flexible which allows assessing the effects of past outcomes on the current one and requires to fit a single model for a stage. The proposed model is exemplified using Health and Retirement Study (HRS) data from the USA.