Logistic Regression Algorithm-Based Product Recommendation System Model

It is very rare to see that the product development process is smoothly carried out with no trials and errors. In the process of producing new products, many difficulties are frequently occurring contrary to the expectation. In order to efficiently cope with such difficulties, it would be required to have the procedure for reflecting consumers’ needs by introducing and applying the recent Opinion Mining, Sentiment Analysis, and Logistic Regression to the product development process. This study used the linear regression and logistic regression models to predict a product with the highest purchase possibility, which was used for the newest marketing technique for the next best action or product recommendation, by analyzing the massive unstructured consumer data of portal sites and online markets, positiveness/negativeness, and predicting/analyzing the causes. Even though the concept itself of those models is simple, they are the most successful models in the logistic regression field. Some people may be opposed to call these models logistic regression. However, the concept itself is the same, and they are achieving such excellent results in the actual logistic regression. The logistic regression model is used for predicting a product with the highest purchase possibility in the newest marketing technique for the next best action or product recommendation. To conquer the new product market, the planning and development of new products are playing huge roles in the corporate performance and growth. The logistic regression-based product recommendation system model proposed by this thesis is the system for acquiring useful and accurate information to plan the best product. In the results of analyzing total 11,200 data sets of a specific product in portal sites and online markets, the customer satisfaction was 92% and there were some product defective issues in case of frequency analysis, so that it would be necessary to have the urgent improvement. According to the results of a survey on the use of the proposed model targeting the planners of products related to system operation, the items related to the system use satisfaction, system efficiency, and system effectiveness showed the satisfactory results higher than expected.