Collaborative clustering increases the researcher’s interest since from last decade for sufficient recommender systems. Thereare different methods presented for CF which is based on two different categories such as user based CF and item based CF. in userbased CF, main task is to find out set of users who have similar favor patterns to a given user and recommend to the user those itemsthat other users in the same set like. In item based CF method, main task is to provide a user with the recommendation on an item basedon the item with high correlations. But exiting methods are having limitations like data sparsity, recommendation accuracy andprediction error. Here, we have proposed novel efficient Expectation Maximization (EM) Clustering and typicality-based collaborativefiltering recommendation (EmTyCo). EmTyCo method is based on existing Tyco method by using efficient Expectation MaximizationAlgorithm for clustering purpose