The best option for a random sample of lamps to be inspected in a hotel with 12,000 lamps across 30 floors is:
100 lamps on each floor chosen randomly
This option ensures that each floor is represented equally in the sample and reduces bias that could arise from only selecting lamps from certain areas of the hotel. The other options either target specific types of rooms or floors without considering an even distribution, which could lead to misleading results about the overall condition of the lamps in the hotel.