Abstract: Dance motion is a rhythmic structure which can be represented by digital pattern of body movement and estimated by mathematic term of vector, matrix and tensor data. Our study proposed estimation method through tensor decomposition and normalization process and increase the efficiency of estimation by combining tensor rule and CP-ALS method. The methods are applied in the detection of Golek Menak motion dance. The tensor data are processed in BioVision Hierarchical (BVH) matrix to estimate the whole structure of the dance motion for normalization process. Our study result showed that the quality of tensor data can be improved after the normalization and the matrix decomposition are processed to estimate the dance motion.