The current methods for selecting time series of cloud services are failed to take into account the user preference, while the traditional user preference algorithms are only applied to the case that QoS indicators values are single numerical. QoS indicators values for time series vector case cannot be effectively quantified. This paper proposed a time series of cloud services selection algorithm which employs subjective and objective weight. The new algorithm could adapt to the different level of user preference with a subjective weight calculation method, and could evaluate correlation with an objective weight calculation method. Experiments show that the algorithm not only avoids the vagueness of user preference, but also accurately reflects the performance of the services. So the results have high accuracy and scientificalness.