This document presents the results obtained in the construction of a predictive model for the yield of corn crops in Colombia to accurately estimate the yield that will be obtained at the end of the harvest, based on a set of climatic variables. Using the CRISP-DM methodology as a framework, a systematic review of literature related to the case study is carried out, in order to identify the essential elements for the research. A climate data set consisting of a historical record of seemingly unrelated data is compiled, and from this, the variables with the greatest influence on performance estimation are determined through statistical analysis, correlation calculations, and relevance assessment. Then, a predictive model is developed with three variants implemented using computational tools and various deep learning (DL) algorithms, with the purpose of identifying the one that offers the best performance through its training and validation. The final model uses sunshine, precipitation, vapor pressure, and maximum and minimum temperatures as input variables, and is based on the DNN algorithm to estimate crop yield. This model provides highly precise results that, compared to those obtained by models found in related research works, stand out for their relevance and accuracy.