A method for indoor localization based on previous activities. The method includes acquiring a target observation from a target user, adding the target observation to a target observations list of a plurality of observations lists, obtaining a target cluster for the target user, verifying that the target cluster satisfies a localization condition, replacing the target cluster with a substitute cluster responsive to the target cluster not satisfying the localization condition, obtaining an estimated location of the target user based on the target cluster responsive to the target cluster satisfying the localization condition, obtaining the estimated location based on the plurality of observations lists responsive to the substitute cluster not satisfying the localization condition, and storing the estimated location in a database. The target observation is acquired utilizing an access point.