Twitter sentiment analysis of Ikea

Sentiment analysis is a natural language processing technique to classify textual data. It is one of the main approaches in machine learning that is used to enhance decision-making by gaining knowledge underlying the growing amount of historical online text data. In this thesis project, the sentiment analysis method was applied to analyze the perception of people on IKEA products. As part of this study overall, 60,000 tweets were collected from the Twitter platform. After predicting the polarity of the tweets, they were divided into categories to investigate the polarity of users on IKEA products. Multiple machine learning methods were used to predict the polarity and RoBERTa had the highest accuracy of 0.70 and f1-score of 0.7023. However, Naive Bayes also performed well, with an accuracy of 0.65 and f1-score of 0.65. On the other hand, traditional machine learning algorithms including KNN and BERT show poor performance. The outcomes of this work provide a prototype on future performance analysis considering user attitudes, perceptions, and opinions on IKEA. Towards this goal, an academic approach was followed by performing exploratory data analysis experiments in order to investigate the reason for frequent user mood changes in social media