Optimized Fused Regression Model for Regression Algorithms

Abstract

One or more independent variables are compared to a dependent variable using regression analysis. Prediction and inference are its key goals. This strategy helps identify data patterns and trends to estimate constant outputs from variables. This research examines how Gradient Adaptive Moment Estimation Optimiser and ensemble multiple linear regression may improve regression task prediction. OFRM efficacy is assessed using six datasets from distinct sectors. six datasets from different domains were utilised to test OFRM. Test it against five regression models. Apply strict criteria and test OFRM extensively to establish its impact on anticipated accuracy, robustness, and generalisability. OFRM dominates individual regression on all datasets. This research shows OFRM's performance in regression scenarios to advance ensemble learning. This paper emphasises the necessity to combine optimisation and ensemble techniques to enhance regression models for real-world applications. Regression model performance on NFT datasets was evaluated using MSE, RMSE, MAE, and R³ measures. OGFR predicts accurately with the lowest MSE (1.04), RMSE (2.21), and MAE (1.29). Best fit is achieved with a R² value of 0.85, accounting for 85% of sample variance. By outperforming DNR and KNN with R² of 0.55 and MSE of 1.91, OGFR is the top model for NFT dataset predictions with a R² of 0.75 and MSE of 1.24.