Enhance Market Price Prediction Accuracy For Second-hand Luxury Goods Through Sentiment Analysis

Abstract

As sentiment expressed in people’s comments can be systematically recorded and amassed in significant volumes, leveraging this information to augment pricing models stands as an effective means to enhance predictive accuracy. The objective of this study is to explore the impact of comment occurrence on price fluctuations and investigate the causal relationship between changes in the value retention rate of luxury goods and fluctuations in public sentiment derived from online reviews across ten prominent luxury brands. Initially, we employ web scraping techniques to gather thousands of comments and gather pricing data from multiple specialized second-hand luxury goods platforms. We then conduct sentiment analysis[1] and utilize ANOVA test[2] to scrutinize whether the appearance of comments significantly influences price variations. Following this, we compute the value retention rate of luxury goods and conduct Granger Causality test[3] between value retention rate and sentiment score. Our hypothesis testing corroborate that public sentiment notably impacts the pricing of second-hand luxury goods, thus advocating for the integration of a new feature, ’sentiment,’ in second-hand luxury goods pricing models. Lastly, we employ machine learning methodologies[4] to develop more refined pricing models and asses the predictive accuracy of these models with and without the sentiment score factor.