 What kind of actions are feasible after we make our predictions? We'll give some examples. The first one is detecting certain patterns that seem to suggest that students are at risk. The type of patterns or situations we refer to are students about to drop out of the course, for example. Another situation in which an intervention could be possible is by monitoring the type of resources that a student search for. That could provide as a position to provide automatic resource recommendation. Another example very interesting is personalized notification. A student might need to be warned that a deadline is approaching. Another student might be warned that should read certain material and so on and so forth. Then there is the immediate feedback that tutors can provide if they have detailed information. That in itself is extremely powerful tool. And finally this type of observation also puts us in the position to detect student emotions and act accordingly. All in all, learning analytics is portrayed as a very promising context to improve the learning scenario.