Sentiment analysis on Zomato customer reviews using Random Forest Classifier

Sentiment analysis which is also referred as Opinion mining is the process of determining emotion behind a text or message. Zomato, a prominent player in food and restaurant industry and has become a best choice for gauging customer’s sentiments and preferences. It hosts a plethora of customer reviews that provide valuable insights into the food experiences of its users. This paper presents a comprehensive analysis of Zomato customer reviews which utilizing a Random Forest classifier to discern sentiments associated with restaurant visits. And its primary objective is to develop an effective sentiment analysis model capable of categorizing customer reviews into positive, negative, or neutral sentiments accurately. In order to gain these objectives, initially, we leverage a dataset from Kaggle Hyderabad restaurants, comprising a diverse range of customer reviews. Then we employ natural language processing (NLP) techniques like TF-IDF to preprocess and extract valuable features from textual data. Followed by, splitting the dataset into training set(70%) and testing set(30%). Subsequently, a Random Forest classifier is trained using the training set, harnessing the power of ensemble learning to enhance sentiment classification accuracy. Finally, the model is evaluated by the metrics(Accuracy, Precision, Recall). The results obtained are average accuracy of 93% and average precision of 93%, recall of 87%, showcases the effectiveness of the Random Forest classifier in accurately categorizing customer sentiments within the context of Zomato reviews. This helps us provide insights into the factors that influence sentiment, helping restaurant owners and managers understand the key drivers behind customer’s contentment. Additionally, this paper explores the implications of sentiment analysis in the context of restaurant recommendations, enabling personalized dining suggestions for users based on their preferences.