The volume of textual data, more specifically, the magnitude of opinionated text on social media, has increased the interest of companies to closely analyze what their customers have to say about them and their products. This thesis explores the possibility of performing aspect-based sentiment analysis with YouTube comments. The comments on Nokia Mobile phones are the subject of the study in this thesis. First, manual labeling was performed to identify the aspect terms and sentiment and then categorize the aspects based on the aspect’s functionality on the phone. From the categorization, it was found out that people mainly have shown negative sentiment towards multiple aspects of the phone with maximum negative attitude towards the price of the phone. On the other hand, the only aspect that could gather a positive attitude was the phone’s-built quality. The result shows that there are multiple phone aspects that HMD Global can consider for current and future product improvement. Further, this study used the labeled data to perform supervised learning to classify the aspects and the aspect sentiment from the comments. With two features extraction techniques, BoW and TF-IDF, this paper has explored the performance of different machine learning models on YouTube comments. The models show good results for aspect classification; however, the model’s performance could be further improved for aspect sentiment classification. Overall, little attention to this area has been discussed because of the complexity, highly unstructured, and noisy nature of text on YouTube. However, despite the challenges, this platform can be valuable in producing insightful analysis, as presented in this thesis.