As the world's population of senior citizens continues to grow, the burdens on the professionals who care for them (carers) are also increasing. In nursing homes, carers need to make a daily report for each resident aiming to improve his/her quality of life. However, in the present understaffed situation, it is difficult and burdensome for carers to record the resident's activity in detail since each carer needs to take care of several residents at the same time. In this paper, we propose an automatic daily report generation system which can monitor the activity of multiple residents in nursing homes. Knowing that important activities such as toilet, bathing, rehabilitation and so on take place in specific areas in a nursing home, it is possible to record residents' activities by tracking their stay areas and movement between the areas within the day. Our proposed system estimates stay areas of multiple residents by machine learning for RSSI values that are sent from BLE beacons attached to residents and received at BLE scanners deployed over multiple areas, and records activities of the residents determined based on their estimated stay areas. The proposed system can also output a daily report of each resident based on the recorded data. We carried out a five-day experiment with four elderly participants in a nursing home and evaluated activity estimation accuracy by leave-one-person-out cross-validation. As a result, our proposed system achieved the weighted average F-measure of 81.6%.