Sometimes outcomes are not equally likely, for example with a biased dice or a drawing pin. Then the probability must be estimated from an experiment. The relative frequency of an event is the fraction of the trials in which it happened.
Relative frequency is only an estimate, and it changes each time the experiment is repeated. The more trials there are, the more reliable the estimate is.
Expected frequency works the other way round. If you know the probability, you can predict how many times the event should happen in a given number of trials. This is an average, not a promise: the real result is usually close to it but not exactly equal.