Analysis of Neural Network Algorithm in Comparison to Multiple Linear Regression and Random Forest Algorithm

Abstract

Regression analysis, a stalwart in statistical methodology, offers a robust framework for predicting outcomes based on historical data. It hinges on the premise that by scrutinizing past input data, one can discern the relationships between independent and dependent variables, enabling the forecasting of final results. In the dynamic landscape of Machine Learning, a multitude of regression techniques exists. Nevertheless, many real-world companies grapple with optimizing their return on investment due to the perplexing task of selecting the most apt model for their specific datasets. This research endeavor seeks to bridge this knowledge gap by conducting a comprehensive comparative analysis of three widely used and highly proficient regression algorithms: Multiple Linear Regression (MLR), Random Forest (RF), and Neural Networks (NNs). MLR offers a simple and interpretable linear model, while RF harnesses ensemble learning to handle complex relationships, and NN s employ intricate, nonlinear modeling capabilities. The study subjects two distinct datasets, Crop Yield, and Cardiovascular Disease, to scrutiny. The former addresses agricultural productivity forecasting, while the latter explores healthcare applications. By evaluating these datasets using the three regression models, the research aims to determine the most suitable model for each dataset's unique characteristics, enabling data-driven decision-making and enhancing the efficacy of regression analysis in practical, real-world scenarios.