This paper examined the impact of geometric irregularities on the intersection capacity at traffic circles. A new empirical capacity relation was proposed to predict the capacity of the traffic circle as a function of geometric elements, exit and circulating flows. Within this scope, firstly, the relationship between the vehicles entering from the minor approach and the circulating flow from the turning movement in the traffic circle was examined based on the K-Means cluster analysis method. The analysis was created in accordance with an exponential relationship between entry and circulating flow. Secondly, two clusters were selected by the partial least squares regression method to improve the model's effectiveness. Lastly, to validate the model, "leave-one-out" cross validation was used to select the components that maximize the model's predictive ability. The results show that geometric parameters of a traffic circle create different effects on capacity, especially in different circulating flow conditions.