The Internet of Things (IoT) and cloud computing have transformed various industries by facilitating intelligent applications and improving data processing. While IoT devices generate massive volumes of data, cloud computing provides scalable resources for storage and analysis. However, integrating IoT with cloud computing poses challenges, such as latency, bandwidth limitations, and resource allocation. Traditional static processing models fail to adequately improve performance in dynamic IoT environments. This paper presents a novel simulation approach combining SimPy for discrete event simulation with a neural network-based decision-making system. Our method optimises data processing techniques across IoT, edge, and cloud environments. We evaluate the efficiency and scalability of centralised, edge, and hybrid data processing models in cloud-IoT scenarios, determining the most effective technique for each use case. The results demonstrate that edge processing can greatly decrease latency for applications that require real-time responsiveness, while hybrid models are particularly effective in IoT contexts that handle large amounts of data. The high precision of the neural network in predicting optimal methods demonstrates its potential to offer a flexible and adaptive framework for optimising IoT-cloud integration, paving the way for more efficient and responsive IoT systems across various sectors, such as smart cities, industrial IoT, and healthcare applications.