With the emergence of Big Data and Cloud Computing, more and more data analytic software services have become available through a Cloud platform. Compared to the traditional service selection problem, selecting this type of services has additional challenges, which requires new selection models being proposed. It is the purpose of this work to “create a testbed” to benefit the research community in this area so that different selection models with consideration of different performance-influencing factors such as algorithms implemented, datasets to be processed, hosting infrastructure, can be tested and compared. We created a cloud-based platform for publishing and invoking analytic services as well as monitoring service performance during invocation. We implemented various data mining algorithms from different packages as example analytic services and hosted them on different infrastructure services. We also ran these services on some real datasets to collect a sample dataset of their Quality of Service (QoS) values.