Abstract
Petroleum price prediction is a challenging task due to the strong volatility in the data. The present study details the development of a predictive model for petroleum prices utilizing two popular algorithms, linear regression and random forest. Based on the analysis of experimental results, it was observed that the model yields an acceptable level of error within the established parameters. Nevertheless, certain limitations were identified, such as the inadequate performance of linear regression in cases where the variation between data points is significant or where the relationship is non-linear. Similarly, random forest algorithms may suffer from reduced accuracy when handling data sets with small sample sizes or low-dimensional data. As a means to address these limitations, the study proposes the application of regularization techniques to mitigate overfitting in linear regression, as well as the handling of null and missing values. Additionally, improvements to random forest performance are proposed, including an increase in the number of trees and the application of hyperparameters to enhance accuracy.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have