/Gov-Tech-Hackathon

AI-based crop recommendation system

Primary LanguageJupyter Notebook

Gov-Tech-Hackathon

AI-based crop recommendation system

Overview

This project uses two external APIs:

  1. mapmyindia - to get district and state names from the latitude and longitude provided
  2. weather - to get humidity, temperature, rainfall using the district provided

It then uses 3 models:

  1. Crop prediction - uses weather api and returns best crop (Random Forest Classification)
  2. Fertiliser prediction - uses weather api and returns fertiliser name (K Nearest Neighbours Classification)
  3. Crop yield prediction - uses crop predicted, state, district, season, area of farm to predict the yield (Gradient Boosted Regression)

Database Usage:

It stores the training dataset for crop yield prediction as well as the fertiliser names and corresponding npk values in two separate collections.
The mongoDB database is used for two purposes:

  1. It uses the training dataset to obtain the label enconding of the user inputs during run-time
  2. Uses the fertiliser name predicted to query the corresponding NPK values