Stress prediction is an important task for the design and analysis of any structure. Finite element analysis and machine learning are widely used to reduce the time, cost, and effort consumed in laboratory experiments. In this paper, multiple machine learning models were built in order to predict the Von Mises stresses at critical locations for any given flange thickness, web thickness, and flange width of an I-Section for a given load condition and compare the accuracies with each other. The machine learning algorithms include linear, polynomial, decision trees, random forest, ADR, gradient boost, bagging, and extra trees, for which the train data was taken from finite element analysis results. Twenty I-section geometries with different dimensions were taken, and analysis was done to create the training set to train the machine learning algorithms. Finally, a comparison between results obtained from finite element analysis and machine learning was made.