In this study, cassava (Manihot species) dewatering centrifuge was modeled as a multivariable machine using dimensional analysis. The efficiency and throughput of the centrifuge were modeled. Also, responses such as product moisture, product recovery, effluent, mass number, processing time number and power consumption number were modeled. The independent and dependent (responses) variables were measured. Their values such as their coefficient of determination, standard and percentage errors were compared. The correlation coefficients for the dependent variables R² = 0.893, 0.999, 0.998, 0.979, 0.906 and 0.987, respectively, are clear evidence of high degrees of correlation between measured and predicted results. The result of residual analysis between measured experimental results and predicted models ranged between −3.3401% and 2.6009%, −0.3043% and 0.3699%, −0.3625% and 0.7631%, −280.33 and 261.53, −284.68 and 279.47, and −240,148 and 260,358. Percentage (%) errors were between −21.65% and 11.36%, −0.40% and 0.48%, −1.87% and 7.23%, −45.35% and 70.1%, −66.76% and 32.86%, and −47.33% and 47.78%. Their values showed no significant differences between measured and predicted results.