<p>Network traffic classification using machine learning (ML) has gained tremendous attention recently. However, many of the methods proposed for network traffic classification have been limited by the low accuracy of the results. Thus, the given paper presents an ensemble approach to network traffic classification, comprising several phases such as data pre-processing, feature extraction, and classification. Initially, the dataset is pre-processed and balanced using Synthetic Minority Over-sampling Technique (SMOTE). It is followed by clustering using K-means for grouping similar data. For the solution, the Random Forest (RF) classifier is used for feature extraction. The random forest (RF) classifier is then applied to the clustered data obtained from K-means clustering, using the features that have been extracted. The voting classifier approach is used, which predicts the outcome according to the class with the highest probability. The performance of the proposed approach is validated based on evaluation metrics such as accuracy, precision, and recall, resulting in a 99.01% success rate, which is higher than existing recent studies in the context of network traffic classification.</p>