Reviewing System using Exploratory Data Analysis and Ensemble Machine Learning Algorithms

In the fast-growing technological advancements, every business venture is thriving for success. Irrespective of the business’s paradigm, each one has a reviewing system that envisages the working throughput of the system. The impact justifies the company’s current commercial state and also helps to bring in new customers. The proposed exploratory data analysis shows the relationship between the reviews on textual and non-textual parameters aimed to work on a user-based analysis. In this paper two different datasets are analyzed where the reviews are predicted based on extensive exploratory data analysis and feature engineering. This paper focused on examining how reviews are influenced based on the independent parameters from Zomato and Yelp datasets. These datasets have huge geographical differences that allowed us to obtain various diverse users-review coming from different backgrounds, cultures, and heritage. After feature construction, the predictive model is analyzed by applying supervised and ensemble machine learning algorithms such as Linear regression, Logistic regression, Decision trees, RandomForest, XGBoost, and GradientBoosting. The inferences obtained from both datasets gave us coherent results that validate the experimental procedures, extend the same analogy in different business paradigms, and offer a comprehensive prediction.