Network equipment receives, from User Equipment (UE), information indicative of each of multiple UE locations in a wireless communication network. For each location, a respective antenna beam direction for communications between the network equipment and a UE at each location is determined. A Machine Learning (ML) module is trained using each location as an ML module input and the respective antenna beam direction for each location as an ML module output. The trained ML module is then used to provide beam predictions based on UE location. Based on the current location of a UE, one or more antenna beam directions for communications between the network equipment and the UE at the current location are obtained from the ML module. The UE location is input to the ML module, and the antenna beam direction(s) are then obtained from the ML module output(s).