Collaborative filtering has received significant attention recently due to practical applications and the popular Netflix Prize competition. Matrix factorization methods have been among the most popular and most successful approaches for this problem. In this work we propose an algorithm for binary principal component analysis (PCA) that scales well to very high dimensional and very sparse data. Binary PCA finds components from data assuming Bernoulli distributions for the observations. The probabilistic approach allows for straightforward treatment of missing values. The method is applied on the Netflix data. In the presentation I will describe the binary PCA method, the experiments and some practical implementation details.