An In-Depth Rainfall Classification Using Random Forest And Artificial Neural Network

In the fisheries sector, accurate rainfall prediction is very important because it affects the condition of the aquatic environment, such as salinity levels, water temperature, and nutrient availability that have a direct impact on fish growth and the success of cultivation. Extreme weather due to high or low rainfall can also pose a risk of damage to fisheries infrastructure, such as ponds and aquaculture ponds. In this study, two weather prediction models, namely Random Forest (RF) and Artificial Neural Network (ANN), were tested to predict rainfall in Australia using a daily weather dataset from 2008 to 2017 covering 145,000 rows of data. The evaluation results showed that Random Forest provided the highest accuracy of 96.27%, while ANN recorded an accuracy of 80.79%. The difference in accuracy between the two models reached 15.48%, indicating that Random Forest is much more optimal in predicting weather conditions, with more stable and accurate performance. Based on these results, Random Forest is proven to be superior to ANN in terms of rainfall prediction accuracy, making it a better and more reliable model for weather prediction applications to support the resilience and efficiency of the fisheries sector in Australia.