Reliability of traditional statistical techniques depends onvalidity of basic assumptions (i.e., normality, homogeneity,linearity, etc.). General linear models could fail in capturing therelationships between the output and explanatory variableswhen those assumptions are violated. Therefore, nonparametric techniques like Classification and Regression Tree(CART), Chi-Square Automatic Interaction Detector (CHAID),Exhaustive CHAID, Multilayer Perceptron (MLP), Radial Basis Function (RBF) and Multivariate Adaptive RegressionSplines (MARS) not requiring distributional assumptions arefavorable. These methods can handle nominal, ordinal andscale variables. CHAID and Exhaustive CHAID algorithmsdiffer only in merging stage and are same in splitting and stopping stages. They can analyze continuous and categorical responses. CART (binary-splitting) and CHAID (multiway splitting) are tree-based data mining algorithms specified only forone response. MLP and RBF are powerful Artificial NeuralNetworks (ANNs) techniques defined for uni- and multiresponses. MARS is one of the most popular data mining algorithms able to solve classification and regression problems. Forregression type problems encountered in agriculture, application of MARS is very scarce. This book contains R commandsfor performing MARS statistical analysis and goodness of fitstatistics using “earth” package (Enhanced Adaptive Regres sion Through Hinges) of R software. The package “earth” wasdeveloped by Milborrow (2018) to perform MARS data miningalgorithm with various penalty options (i.e., -1, 2, 3 and 4) in Rsoftware. MARS model goodness of fit criteria are GCV (Generalized cross-validation), r (Pearson correlation coefficientbetween actual and predicted values of a response variable),coefficient of determination (R2), adjusted coefficient of determination (Adj.R2), standard deviation ratio (SDRATIO), root ofmean square error (RMSE), mean error (ME), global relativeapproximation error (RAE), mean absolute deviation (MAD),mean absolute percentage error (MAPE), Akaike’s informationcriterion (AIC) and corrected Akaike’s information criterion(AICc). This book illustrates how to perform MARS algorithmand to compute goodness of fit criteria for some sample datasets at different fields of agricultural sciences. It highlightsimportant points on selecting suitable interaction order andpreventing model overfitting. Some recommendations are provided to obtain a smaller MARS model than constructed bydefault, like how to specify penalty= -1, nprune and nk arguments for a continuous output. Additionally, brief informationon MARS algorithm R commands for executing multiple continuous responses models on the same basis functions is provided in Appendix section. The book intends to introduce researchers in agricultural field with the MARS algorithm byproviding detailed information on its application in R software.