In robot-assisted minimally invasive surgery (RAMIS), optimal placement of the surgical robot's base is crucial for successful surgery. Improper placement can hinder performance due to manipulator limitations and inaccessible workspaces. Traditionally, trained medical staff rely on experience for base placement, but this approach lacks objectivity. This paper proposes a novel method to determine the optimal base pose based on the individual surgeon's working pattern. The proposed method analyzes recorded end-effector poses using machine-learning based clustering technique to identify key positions and orientations preferred by the surgeon. To address joint limits and singularities problems, we introduce two scoring metrics: joint margin score and manipulability score. We then train a multi-layer perceptron (MLP) regressor to predict the optimal base pose based on these scores. Evaluation in a simulated environment using the da Vinci Research Kit (dVRK) showed unique base pose-score maps for four volunteers, highlighting the individuality of working patterns. After conducting tests on the base poses identified using the proposed method, we confirmed that they have a score approximately 28.2\% higher than when the robots were placed randomly, with respect to the score we defined. This emphasizes the need for operator-specific optimization in RAMIS base placement.