Detecting low-dimensional chaos in time series of finite length generated from discrete parameter processes