Some people classify room colors into base color, assorted color and accent color when they think about coordination of room. The proportion of the classified colors hereinafter called color region ratio. Studies have been conducted on the impression that the color of a room gives humans and on the color preferences of the room. However, the color ratio of the room has not been sufficient studied. Therefore, we suggest a method for room evaluation that uses the color region ratio of the room, the strength of the complementary colors and the length of the boundaries which is lines between the different colors after clustering. In order to provide highly accurate room evaluation with less data, we use ensemble learning from the examinees’ room preferences. The average correct response rate for training data is \(94.8\%\) , and unknown data is \(87.5\%\) . Thus, it is possible to evaluate the room according to the examinees’ preferences.