Co-location pattern mining plays an important role in spatial data mining. With the rapid growth of spatial datasets, the usefulness of co-location patterns is strongly limited by the huge amount of the discovered patterns. To overcome this drawback, several statistics-based methods have been proposed to reduce the number of discovered co-location patterns. However, these methods cannot guarantee that their mined patterns are really user-preferred. Therefore, it is crucial to help users discover the co-location patterns they preferred through effective interactions. This paper proposes a newly interactive approach based on SVM, in order to discover user-preferred co-location patterns. First, we presented an originally interactive framework to help the user discover his/her preferred co-location patterns. Then, we designed a filtering algorithm with a small part of patterns as the training set of the SVM model for the provision of each interactive process, by which a high efficiency could be achieved by the optimization of the SVM model. Finally, the system was verified on both the real data sets and the synthetic data sets, accompanying with the 80% prediction accuracy.