How should I pass clustered dataset to Apache Mahout recommender?

Question

I am working on Movielens dataset and I want to create a user based recommender system which is based on k-means clustering. I have created the clustered dataset using Python and now I want to use it in Mahout for better recommendations.Finally, I have to analyse the effect of k(no. of clusters) on the MAE and RMSE values.Any help will be appreciated, thank you.