Agrogen uses machine learning to forecast produce prices and help U.S. farmers decide when to sell. The platform runs on Google Cloud and shows price predictions for the next three days in best and worst case scenarios, accounting for inflation, weather, and location.
In the United States, farmers face an array of problems such as fluctuating produce prices, unsustainable income, and extreme weather. These factors can lead to economic insecurity and higher debt for those small farmers in the farming community. Introducing Agrogen, a one-stop cloud solution that uses machine learning to forecast produce prices and advises farmers about when to sell. The model considers factors such as inflation, weather, and location, and combines it with a simple user interface for the farmer to use. Agrogen uses Google Cloud, MySQL, Flask, and Python on the backend. The Flask app, app.py, runs the database as well as the models. It provides API routes to access all the MySQL queries as well as a compiled version of the React frontend. On the frontend, we use React and Tailwind CSS in order to style components and save state between pages. Axios is used to interact with the backend APIs in order to perform actions such as registration, login, or adding produce. On this main page, we give some basic information about the site and encourage them to register. So a farmer will visit our site and they may decide to register. And if everything goes correctly, they should now be registered for the site and should now be able to sign in. So as you can see there, I put in the wrong username and it returned an error. So now that I'm signed in, you'll notice that the menu bar changed. It can— I can now access my inventory and my predictions. Since we don't have any produce, we'll visit that page first. We first add produce. We'll be able to select which produce we want to get predictions for. As well as the region in which our crops are stored in order to ensure the model is more accurate. And we will add all the produce that we need. And if we are now no longer interested in tracking that produce, we can remove it from the list. So on the backend, all of this is being reflected in the MySQL database. So after we're done adding produce, we'll then be able to view our projections. We navigate to this page, it loads the pre-trained model and runs predictions for today as well as the next 3 days and displays the 3 previous days of data. This allows the farmer to get an accurate picture of the trend of the crop for the week. We provide 2 models, a worst case and best case scenario for the crop that week. Thank you.
