This research develops mathematical models of C, M, Y, K and CIE L^∗a^∗b^∗ chromaticity values to minimize the errors between the predicted ratio and the actual ratio of process colors (cyan, magenta, yellow, black) on holographic paper by measuring and analyzing the data after mixing the basic inks according to proportion. It obtains the least squares estimation points by performing the multiple nonlinear regression analysis method in MATLAB. Moreover, by replacing the values of CIE L^∗a^∗b^∗ chromaticity, the regression significance of the mathematical model is verified and the corresponding basic ink mass ratio is obtained. The results reveal that the RMSE, MAE, and R² values of spectral prediction models, which are established by multiple nonlinear regression analysis, show small errors between the predicted and the actual outcomes.