COMPARISON OF MACHINE LEARNING ALGORITHMS FOR PREDICTING DIAMOND PRICES BASED ON EXPLORATORY DATA ANALYSIS

Abstract

Diamonds are a unique commodity whose socially generated notions significantly influence perceived value. To study how a diamond's physical attributes might predict its price, a massive dataset of loose diamonds scraped from an online diamond store is subjected to data mining, which reveals that diamond weight, color, and clarity are the most influential determinants of diamond pricing. Therefore, submit a proposal for an Exploratory Data Analysis that includes a component that analyses various parts of news articles using LASSO Regression, ElasticNet Regression, and Random Forest Regression. This system is trained on past data to forecast diamond prices while retaining an easily interpretable trading approach concerning rule complexity. The suggested strategy beats cutting-edge methods for prediction accuracy and interpretability, such as extreme learning machines using deep learning. Our data indicate that the news impact factor is crucial for forecasting. Demonstrate that the suggested system outperforms the average yearly return while offering a set of language trading rules that are interpretable. This has substantial repercussions for investors. A significant degree of subjectivity in diamond pricing may result from diamond dealers' price concealment techniques.