Neural networks are a tool that allows us mathematically model a process, whose model based on physical principles is truly complex. The present study uses this kind of empirical modeling as a tool to predict what will be the production of the biodigester of the pilot plant from Universidad de Piura using cocoa pod husk as biomass. Experimental data were obtained for 58 days that served for the training and validation of the neural network. This study has considered the influence of parameters as weight of biomass, volatile solids, total solids, pH and temperature and can predict the production of biogas, methane and carbon dioxide.