Computer scientists and physicists have paid increasing attention on the link prediction in complex networks recently.The purpose of link prediction is to estimate whether a link exists between two nodes based on the known links and the properties of the nodes.After transforming time series into complex networks, time series prediction can be partly regarded as the problem of link predictions in complex networks.In this research, we discussed how to predict time series with the aid of the topological features of complex networks induced from time series.Our novel predicting framework is helpful to understand the relationship between the dynamics of the time series and topological structures of complex networks induced from time series.And we found that links in complex networks induced from time series are easier to predict than the links in other types of complex networks.