Cognitive radio networks can achieve high spectrum efficiency through dynamic spectrum allocation to the secondary users (SU) by way of using primary user (PU) spectrum. An efficient handoff scheme for cognitive radio networks using hybrid priority queuing model has been proposed using discretion rule. This method is divided into two parts: (1) hybrid priority queuing model with discretion rule is proposed by prioritizing the secondary users in the channel and then calculate the waiting time during the spectrum handoff and (2) to obtain better efficiency by reducing the waiting time, the learning rate of the secondary users from the other users should be increased. This can be done with multi teacher apprentice learning (MAL) along with sparse regression by considering multiple users into consideration at a time instead of a single user for better efficiency of the spectrum. The normalized load, arrival rate and the service time of the proposed model is compared to the existing models with respect to the average data delivery time of the users. © 2019, Springer Science+Business Media, LLC, part of Springer Nature.