The biproportional factorial analysis

Beyond the Factorial Analysis of Correspondences, the paper presents a newmethod of data analysis: the Biproportional Factorial Analysis. In the Factorial Analysis of Correspondences, the matrix to be diagonalised is the product o f the two matrices o f profiles, row and columns: this matrix is not symmetrical. In the Biproportional Factorial Analysis, the matrix to be diagonalised is the symmetrical product o f an intermediate matrix over itself; this intermediate matrix is calculated as the biproportion o f the data matrix over normalisedmargins. This provides a full symmetry between rows and columns. After recalling the Factorial Analysis of Correspondences, the paper recall what it is b…