About

We decided to use U.S. states' weather data (1975-2024) and the annual corn yields by state. We then sorted these states into different regions:

  1. Humid coastal cold
  2. Humid inland cold
  3. Humid coastal hot
  4. Humid inland hot
  5. Dry inland hot
  6. Dry inland cold
  7. Dry coastal cold
  8. Dry coastal hot
We loaded and cleaned all of our weather and corn yield data in R. Since we pulled the datasets from many different sources, we condensed the data into state-specific dataframes. We implemented a deep neural network (DNN) to find relationships between our weather data, regions, and crop yields, and make corn yield predictions.

Why does this matter?

There is a very real issue of uncertain crop yields as a result of increasing average temperatures around the world. Additionally, the World Health Organization estimated that with the increasing World population, the amount of large scale cereal grain (corn, wheat, rice, etc.) production will need to almost double. Our project was an attempt at understanding a possible solution, or rather framework, to help address this uncertainty and concern. Our concept could very easily be expanded to include other crops, and with only requiring temperature and precipitation data, we make this technique accessible to even the most disadvantaged groups.