Diamond Prices Estimation with ELM-Based Regressors

Society uses precious gems like diamonds for many specific purposes, but it has always been complex to measure their value. For this reason, we will use the Extreme Learning Machine (ELM) to estimate diamond prices using a popular database, which contains 53,943 round-cut diamonds. The results show an optimal accuracy for both standard and regularized ELM regression models. The metrics studied in this work include Mean Squared Error (MSE), correlation coefficient, and, in terms of complexity, the number of neurons in the hidden layer and training times. The results showed low Mean Squared Error and correlation coefficients between 98% and 99% in most studied cases, similar to those seen in the state of the art. Our most representative metric for selecting the most optimal method is training time. We conclude that the regularized ELM using database No. 1 presents results with shorter training times.