Data Driven Approach in Building Consumer-Friendly Business Models Using Machine Learning Techniques

This study is about analyzing real-time data sets regarding customer issues, sub-issues, complaint received and company response through machine learning techniques. The central aim is to generate the effective prediction model to resolve the company customer satisfactory model. Model prediction is performed on supervised logistic regression, random forest and extreme gradient boosting techniques based on the data complexity. Performance metric like accuracy, F1 score, R2 value and root mean square error are observed for better accuracy of the model. Train and test case metrics are observed for better predictive model. Among all three mentioned models it is observed that Random Forest is performing better from their metrics like accuracy of 81.61%, -0.104 R2 values, 0.482 root mean square error and 0.280 F1 score values which almost near to train case 81.63, -0.098, 0.482,0.289 respectively. The proposed models are performance is acceptable range due to its complexity of the data set. More powerful techniques with better computational facilities is need of the hour for future Artificial intelligence applications.