Smart Agriculture in the Era of Big Data: IoT-Assisted Pest Forecasting and Resource Optimization for Sustainable Farming

The goal was to facilitate quick development and revitalisation of rural areas by utilising IoT knowledge to accomplish smart agriculture against the backdrop of big data. Depleted soil fertility, increased pest attacks, and diseases are some of the climate-related risks that the agricultural sector is battling. Predicting Utilising directly sensed environmental conditions as input, a machine learning model incorporating Random Forest and the Catboost algorithm determines, based on the current environmental conditions, whether pest populations are present or not. Next, Hybrid Weighted Ant Colony Optimisation (HW ACO) is used to fine-tune the Catboost algorithm. This method evaluates the machine learning model's accuracy and precision in making predictions using test data. Predictions have been made using data from 2028 to 2022, a span of five years. Weekly forecasts are produced by the pest prediction model. With high F-measure, Precision, and Recall, the weekly predictions have an overall accuracy of 93%. Over time, the precision of the pest forecast steadily rises. Predictions for the next week are made by averaging all weather data from the past seven days. For immediate defences against pests, the weekly forecasts are crucial. The findings support sustainable farming methods and food security by showing that the device can maximise crop yields while decreasing resource consumption.