Advances in sensing technology have generated multimodal datasets with complementary information in the agricultural industry. These datasets can potentially be used for developing accurate predictive models. For example, genotype and phenotype (e.g., plant height and canopy size and leaf area) data, hyperspectral images obtained by drones, and soil and weather data can potentially be integrated to accurately predict the yield of a crop. In this project, we will take advantage of dimensionality reduction techniques to extract features from different data sources to be integrated for crop yield prediction. Specific crops that we consider are wheat and coffee beans.