The study aimed to determine the heating values (HV) and proximate composition of the four varieties of Napier Grass (Pennisetum purpureum Schumach) and native wild sugarcane (Sacharrum spontaneum L.)  to generate a mathematical model for predicting HV of Napier Grass varieties and wild sugarcane biomass. The four varieties of Napier Grass namely King, Florida, Dwarf, and Princess Caroline as well as the wild sugarcane were used in this study. The proximate composition such as moisture, ash, organic matter (OM), carbon (C), nitrogen (N), C/N ratio, and heating value (HV) were determined to generate mathematical models. Only ash, nitrogen, organic matter (OM) and carbon (C) have statistically significant r values toward the heating value. Positive correlation was obtained in OM and C whereas negative correlations were obtained in ash and N. Linear regression using Waikato Environment for Knowledge Analysis (WEKA) machine learning generated mathematical models for group variables for predicting the heating value (HV). The organic matter (OM) content showed the most accurate model among the group variables used with adjusted R2 value of 0.900, Pearson’s r value of 0.915 and root mean square error (RMSE) value of 481.64 MJ/kg. Among the proximate composition parameters tested, OM is the most accurate predictor of the HV of the Napier Grass varieties and wild sugarcane biomass.