Heat transfer flow of non-Newtonian eyring-powell fluid with mixed convection heterogeneous and homogeneous reactions using linear regression based machine learning approach

This research examines the heat transfer and flow behavior of non-Newtonian Eyring-Powell fluids under mixed convection and reactive conditions employing a supervised machine learning-based linear regression methodology. The most appropriate transformations are utilized to convert non-linear coupled partial differential equations into an ordinary differential equations. The model predicts thermal behavior on velocity, energy, and concentration boundary layers for four scenarios. To validate the reliability and accuracy of the supervised machine learning model, a comparison and performance evaluation are conducted by using Mean Squared Error (MSE), error histograms, and fitness curves. The average squared difference (MSE) between the observed and predicted values is calculated for the training, validation, and test sets. An error histogram displays the distribution of prediction errors across different sets to aid in identifying biases or inconsistent model performance. It is seen that the higher model performance appears by lower MSE values. It is also observed that the differences in predictions between machine learning and MATLAB methods indicate that machine learning (ML) can better handle a large variety of complex data. It is also concluded that the machine learning method surpass traditional computational techniques, especially in accuracy and efficiency.