Some algorithm like naive bayes and Decision tree works on labeled data where you have a classification column. For example if you want to relate the weather status and day of week with the punctuality of train then it should be labeled data. Because you have lot of combination of weather and day of week and you have a class column containing values late/not late.
On the other hand Apriori Algorithm works on unlabeled data where you don't have class column. For example if a customer buy A and B then most probably he will buy C. Here there is no class column. Any item can come in the decision. This algorithm is used to find association rule and mainly find the frequent item sets from the data sets,