“…Regarding the domain of software project prediction, we found two studies that use MARS along with other techniques to find the best method for predicting software effort [49,50]. • Some studies compared the performance of MARS with that of linear discriminant analysis [34,35,45], neural networks [34, 35, 44-46, 49, 50], support vector machines [35,46,49,50], tree-based approaches [35,49,50], linear regression [46,49,50], logistic regression [35,48], and ARIMA [46]. • Two studies concluded that combining MARS with another technique outperformed the rest of the methods, such as MARS with a neural network [34] and a hybrid fuzzy clustering-MARS [45].…”