Data uncertainty can be generated for a variety of reasons, including measurement error, sampling error, environmental monitoring, sensor networks, and medical diagnostics. The process of mining information from emerging applications such as sensors or location-based services should be handled carefully to prevent erroneous outcomes. Several heuristic techniques and Machine Learning (ML) techniques have been used to classify data in the presence of uncertainty. This paper proposes a novel ML technique by combining the Radial Basis Function (RBF) network with the particle-swarm optimization algorithm. We begin by applying a significant level of uncertainty to a data set and then cleaning up training data and applying data normalization. Next, an RBF network is trained by the optimizer algorithm of particle swarm. Finally, we compare our proposed method with well-known ML techniques namely, k-nearest neighbor (k-NN), Support Vector Machine (SVM), Random Forest (RF), Decision Tree (DT), Logistic Regression (LR), and Extreme Gradient Boosting (Xgboost). Our model outperforms previous standard and well-known ML approaches as it has the lowest error rate according to error metrics. The results of the experiments show that our proposed method is better at classifying uncertain data than previous methods based on standard performance metrics.

Uncertain data, Machine Learning, Radial Basis Function, Particle Swarm Optimization, IoT Applications.