The problem Farm Buddy solves

Farmers face problems daily in choosing the right crop to sow for the next season, preparing for weather hazards, and many more.
The project presents an all-in-one solution for farmers to get crop recommendations for their soil type, get weather warnings through SMS, see current weather predictions, and facility to manage their crops among other features.

We have managed to find the solutions to some of the most challenging things to the farmers like what crops to sow and when to sell their crops for the highest profit. By training machine learning models we predict the best crop suited for the conditions of the farmer's land. This allows them to grow the most optimal crop on their land helping increase their profits. The average MSRP, minimum MSRP, and maximum MSRP is also shown to help him decide the best crop to shown

Challenges we ran into

Subscriptions:

We were using graphql graphene subscriptions to fetch execute tasks asynchronously on our cluster workers and broadcast the result back to the user. The motive of this step was to make the project suitable for production where asynchronous tasks are preferred and the background tasks are executed by other workers. This will help to scale the app as required in the future just by increasing the number of workers.

Formatting data from Government APIs:

The government APIs returned the data in a form that was not friendly to use and present to the user. This required heavy processing and tinkering to extract the useful bits of information from the data received from their APIs

Webscraping Weather Government Page:

We wanted to provide farmers with notifications of adverse weather events, but there were no APIs that would give us the required data to process and warn the users through SMS. So we had to web scrape the meteorological department's weather map data.

Unavailability of drought data:

We wanted to train a model to help prepare for droughts but due to very less or no availability of data ultimately we had to drop this feature.