The only way to do this is via a filter. As Prakhar suggests, you should create a filter and load new data to your already loaded dataset with this filter assigned. Then you can execute a score job against the filter. One thing to be aware of is that your new data should also contain a value for your goal fields. The reason for this is that we've noticed Analytics will degrade your prediction model if you don't. For example, let's say one of your goals is failure rate for a specific component. You've trained your model and Analytics tells you that for condition A your failure rate is 3%, condition B=25%, condition C=46%, condition D=73%. Your training data explicitly defined a failure of 100% or 0% based on all the other data for a given row. Now you start submitting data but you don't specify a failure value for your goal. Analytics looks at the data your sending and sees, based on historical trends, the goal column should indicate some sort of failure percentage. Unfortunately, by not including a value, you essentially told it "0% failure", so it negatively affects the score value that's returned and your model begins to degrade.
Before you submit your new data, you need to query for the prediction results of your trained model for your various goals, compare your new data against the ranges of data returned by Analytics and use that score value for your goal. For example, if attribute A has a value of 44 and the ranges returned by Analytics when you ask for the predictive results of your goal shows that for a range of 42-46 your predictive score is 46% you need to set the value of your goal in your new data to .46. Then Analytics will return a correct score value.
