A Hybrid Approach to Climate Prediction: Neural Networks and Fuzzy Logic

The use of the neural networks, fuzzy logic, and ANFIS (Artificial Neuro-Fuzzy Inference System) integrated models to examine the forecasting of climate condition is the objective of the research study. The aim is to enhance the accuracy of climate forecasting and bring insightful data for decision-makers who will have to deal with the problems of climate change through the evaluation of present computer models. The article is built on the basis of the FNN architecture for weather classification, the dataset of which was downloaded from Kaggle and consisted of eleven classifications of 6862 weather photos. Moreover, the ANFIS (which is a blend of fuzzy logic and neural networks for automatically generating rules) is particular in that it facilitates the fuzzy-rule based classification. The research employs the checks and verifications of the FNN and ANFIS models through assessing their accuracy, precision, recall and F1-score values. The merged model significantly outperforms the individual models, as evident in the high 94% accuracy, 96% precision, 92% recall, and 94% f1-score results. The integrated model is an indispensable instrument in policy-making and adaptation measures definition in the context of climate change, as evidenced by its exceptional talent in tracking climate variations, as emphasized by its 0.92 CRSI. The results showed the central role of the combined use of neural networks and the fuzzy logic methods in climate prediction, hence the need for further research to increase the models’ scalability and tackle complex climate dynamics corollary.