This study aims to obtain optimal configuration data for calibrating parameters of robots in order to enhance the positioning accuracy. Analysis of calibration parameters finds a huge deviation in each coordinate dimension of robot configuration when uniaxial or biaxial robot pose data were used to identify the geometric parameters through the calibration model. Hence, triaxial coordinate data are used. Experimental results show that pose data from three-dimensional planes can yield better calibration precision with the maximum deviation below 0.4mm for any axis. In addition, the minimum number of configuration data sets needed for achieving steady deviation in positioning accuracy is also determined.