This paper proposes an interpretable random forest for opinion mining on hotel reviews. This model performs the task of sentiment polarity about hotel as positive or negative. In addition, we constructed the criteria importance measures to explain and clarify the relationship and interactions of the hotel name, hotel aspect, hotel reviewer, and reviewer time affects the orientation of sentiment is negative or positive. An interpretable random forest was evaluated on three scenarios, which we built based on important features of the hotel reviews. The experimental results on the hotel reviews data set have shown the effectiveness of the proposed issues.