The evolution of e-commas made reviews a crucial metric for judging the quality of online products or services. These reviews have a significant impact on the decision of the customer. Positive review catches more attraction while negative reviews impact sales of the product. Nowadays, deceptive reviews are being deliberately posted on e-commerce websites and social media stores to promote the product by illegal means. These reviews are sometimes posted in different local languages to build a fake virtual reputation among local customers. Thus, fake review detection is a wider area for ongoing research. This paper proposes several machine-learning approaches for the detection of fake reviews written in Roman Urdu. Furthermore, a comparative analysis of the performance of nine machine learning models on the given dataset is performed. The dataset is crawled from different e-commerce sites in Pakistan. The results show that the existing Support Vector Machine outperforms the rest of the models with an accuracy of 82%. This emphasized the critical role of robust detection methods in ensuring the authenticity of online reviews.