A COMPARATIVE STUDY OF MACHINE LEARNING ALGORITHMS FOR GLOBAL SALES FORECASTING

AbstractIn today’s world, Video game industry has developed immensely that it fascinatespeople to a great extent. It is proved to be a noticeable and important contributor to the entireworld in terms of revenue. This field has a greater influence, without any doubt on thepopulation. It has extraordinarily attracted many people with sharp and creative skills to expandand enlarge Video games globally. Discussing about the vast revenue generated by the gamingindustry, Machine learning technologies is widely used to develop models that is highlyeffective in examining and anticipating the sales of the computer games well in advance.Machine Learning mechanism comes up with a good deal of models to envision the future saleswith the help of Linear, Multiple Regression, Random Forest, Decision Trees, Support VectorMachines and more. While all these handles the data by using different mathematical conceptsand formulae to estimate the sales, any or the best model could be selected for the appropriatedata by comparing their accuracy and their performance. The model accuracy measure is takenby total number of correct predictions evaluated by the respective model in comparison withall the predictions. Hence, r-square is extensively used as a performance metric to grade theworking of the model. In particular, four algorithms are tested upon the chosen dataset andfinally compared to detect the finest working model.