Multivariate Adaptive Regression Splines Model Research Articles

Improving the accuracy of satellite and reanalysis precipitation data by their ensemble usage

This study evaluated the outputs of five precipitation (PCP) datasets. These models are ECMWF reanalysis 5th generation (ERA5), precipitation estimation from remotely sensed information using artificial neural networks-climate data record (PERSIANN-CDR), Asian precipitation-highly resolved observational data integration toward evaluation (APHRODITE), The national centers for environmental prediction climate forecast system reanalysis (NCEP CFSR) and climatic research unit (CRU). The PCP outputs of these models were compared with data of nine synoptic stations in the Khuzestan province. The results indicated a better match between the APHRODITE outputs and the PCP data at most stations (R2 > 0.85, root-mean-square error (RMSE) < 17.049 mm and − 4.25 < Bias < 2.633 mm). However, CRU model has the highest critical success index (more than 0.711) and the lowest false alarm ratio (less than 0.2) and ERA5 has the highest probability of detection (more than 0.967) at most stations. Then, PCP outputs of five reanalysis (ERA5), interpolated (APHRODITE, NCEP CFSR and CRU) and satellite (PERSIANN-CDR) PCP datasets were combined to reduce the PCP estimation error. The multivariate adaptive regression splines models were employed for this purpose. The results show that the RMSE of all the stations, except Ahvaz station, decreased and the BIAS decreased too. Given the results, using ensemble data methods is a suitable way for reducing the error and increasing the accuracy of these models.

Assessment of liquefaction-induced lateral spread using soft computing approaches

Lateral spread is a common ground deformation induced by soil liquefaction, which has caused extensive damage to buildings and lifelines in past earthquakes. At present, several empirical methods for assessing lateral displacement have been proposed. Due to the complexity of the problem, it is difficult for traditional methods to predict the lateral spread displacement with high accuracy. In this study, a new database of 522 lateral spreading case histories was established based on the databases developed by Youd et al., Chu et al., and NGL (Next Generation Liquefaction). The confidence ellipse was used to evaluate the similarity between parameters of liquefiable soil layers for two site conditions (free face and gently sloping ground). Then, based on the new database, predictive models for assessing liquefaction-induced lateral displacement were developed by different soft computing approaches including Ridge Regression, Lasso & Lasso CV, Random Forest Regression (RF), eXtreme Gradient Boost (XGBoost), and Multivariate Adaptive Regression Splines (MARS). The MARS model showed good predictive performance (the R2 values were 0.583 and 0.555 on testing set, respectively) and can generate explicit expressions. The results of SHAP investigation showed that log(T15) is the most important feature variable in RF model, where T15 is the thickness of liquefiable soil layer. Further analysis showed that thicker liquefiable soil layers, smaller epicentral distances, larger magnitudes, and unfavorable topographical conditions will increase the possibility of lateral displacement, while larger grain sizes and more fines contents will reduce the probability of lateral displacement.

Using multivariate adaptive regression splines and classification and regression tree data mining algorithms to predict body weight of Nguni cows

ABSTRACT The study was performed to determine the association between body weight and biometric traits and to examine the effect of biometric traits on the live body weight of Nguni cows using Multivariate Adaptive Regression Splines (MARS) and Classification and Regression Tree (CART) data mining algorithms. In total, 105 Nguni cows aged three to four years were used for body weight (BW) and biometric traits viz; head width (HW), head length (HL), ear length (EL), body length (BL), rump height (RH), withers height (WH), sternum height (SH), body depth (BD), bicoastal diameter (BCD), rump width (RW), rump length (RL) and heart girth (HG). Coefficient of determination (R2), adjusted coefficient of determination (Adj.R2), root-mean square error (RMSE), standard deviation ratio (SD ratio) and Pearson correlation between actual and predicted values were predicted to find out the best fit models. MARS models in prediction of BW presented as the best fit as compare with CART model with higher R2 = 0.993 and Adj.R2 = 0.991 with the lowest RMSE = 5.97 and SD ratio = 0.081. In this study, MARS models established are the significant statistical tools that might be used for describing studied breed standards for breeding purposes.

Prediction of Temperature and Carbon Concentration in Oxygen Steelmaking by Machine Learning: A Comparative Study

The basic oxygen steelmaking process (BOS) faces the issue of the absence of information about the melt temperature and the carbon concentration in the melt. Although deterministic models for predicting steelmaking process variables are being developed in metallurgical research, machine-learning models can model the nonlinearities of process variables and provide a good estimate of the target process variables. In this paper, five machine learning methods were applied to predict the temperature and carbon concentration in the melt at the endpoint of BOS. Multivariate adaptive regression splines (MARS), support-vector regression (SVR), neural network (NN), k-nearest neighbors (k-NN), and random forest (RF) methods were compared. Machine modeling was based on static and dynamic observations from many melts. In predicting from dynamic melting data, a method of pairing static and dynamic data to create a training set was proposed. In addition, this approach has been found to predict the dynamic behavior of temperature and carbon during melting. The results showed that the piecewise-cubic MARS model achieved the best prediction performance for temperature in testing on static and dynamic data. On the other hand, carbon predictions by machine models trained on joined static and dynamic data were more powerful. In the case of predictions from dynamic data, the best results were obtained by the k-NN-based model, i.e., carbon, and the piecewise-linear MARS model in the case of temperature. In contrast, the neural network recorded the lowest prediction performance in more tests.

A comparative analysis of biomass torrefaction severity index prediction from machine learning

Machine learning (ML) is one type of artificial intelligence (AI) commonly used for computer programming. Multivariate adaptive regression splines (MARS) and artificial neural networks (ANN) are two common and popular tools in AI that allow the user to analyze the pattern of complex data. The torrefaction severity index (TSI) is an index to define torrefied biomass quality at different torrefaction conditions. In this study, MARS and ANN models are applied to predict TSI. The considered input parameters in predictions using MARS and ANN approaches comprise feedstock type, temperature, and duration. The MARS model indicates that temperature is the most influential factor on TSI, followed by duration and feedstock type. In contrast, the ANN model reveals that the feedstock type is a dominant factor, and temperature and duration are not important. The performance of the ANN model is evaluated in three different combinations of numbers of hidden layers and neurons. It shows 2 hidden layers along with 85 neurons giving the best performance. The highest R2 values in MARS and ANN are 0.9851 and 0.9784, respectively. The relative root means square error analysis shows that both MARS and ANN have good fit quality with the relative errors of 1.49% and 2.16%, respectively. Overall, the comparison reflects that MARS is a more suitable model for predicting solid biofuel’s TSI. The general observation suggests that the ANN lacks sensitivity to the input parameter. Nevertheless, ANN performance may be improved by adjusting the number of hidden layers and neurons.

Undrained stability of pit-in-pit braced excavations under hydraulic uplift

Pit-in-pit (PIP) excavations in an aquifer–aquitard system likely undergo catastrophic failures under the hydraulic uplift, the associated undrained stability problem, however, has not been well analyzed in the past. To this end, a hypothetical model of PIP braced excavation in typical soil layers of Shanghai, China is developed using the finite element limit analysis (FELA) tool. The FELA solutions of safety factors (FSs) against hydraulic uplift are verified with the results from the finite element analysis with strength reduction technique (SRFEA) and existing design approaches. Subsequently, FELA is employed to identify the triggering and failure mechanisms of PIP braced excavations subjected to hydraulic uplift. A series of parametric studies considering the various geometric configurations of the PIP excavation, undrained shear strengths of aquitard, and artesian pressures are carried out. The sensitivities of relevant design parameters are further assessed using a multivariate adaptive regression splines (MARS) model that is capable of accurately capturing the nonlinear relationships between a set of input variables and output variables in multi-dimensions. A MARS-based design equation used for predicting FS is finally presented using the artificial dataset from FELA for practical design uses.

Determining Seismic Bearing Capacity of Footings Embedded in Cohesive Soil Slopes Using Multivariate Adaptive Regression Splines

Seismic bearing capacity of strip footings in cohesive soil slopes considering various embedded depths is investigated in this study. Novel solutions using pseudo-static method and finite element limit analysis (FELA) with upper bound (LB) and lower bound (LB) theorems are presented. The influences of footing depth, slope angle, slope height, undrained shear strength and pseudo-static acceleration on bearing capacity and failure mechanisms are examined using dimensionless parameters. With the comprehensive numerical results, the multivariate adaptive regression splines (MARS) model is then utilized to simulate the sensitivity of all dimensionless input parameters (i.e., the normalized depth of footing D/B, the normalized slope height H/B, the normalized distance from top slope to edge of the footing L/B, slope angle β, the strength ratio cu/γB, and the pseudo-static acceleration factor, kh). The degree of influence of each design parameter is produced, and an empirical equation for the dimensionless output parameter (i.e., bearing capacity factor Nc) is proposed. The study results are accessible in the design charts, tables, empirical equation for design practitioners.

Comparative analysis of soft computing techniques in predicting the compressive and tensile strength of seashell containing concrete

Despite the advantages of using seashells in concrete, predictive models have not yet been proposed for this type of concrete. To fill this gap, the present study utilized three distinctive machine learning models (i.e. Artificial neural network (ANN), Multivariate Adaptive Regression Spline (MARS), and M5P Model Tree) to investigate the compressive and tensile strength of seashell containing concrete using records collected from the technical literature. The results indicate that the ANN model outperforms the other developed models with R2 equal to 0.939 and 0.903 for the test dataset of compressive and tensile strength, respectively. Although the developed MARS and M5P models resulted in lower accuracy, they produced reliable predictions. Thus, the proposed models can be used successfully to estimate the mechanical properties of seashell-containing concrete. Also, sensitivity analysis of the best model follows experimental results highlighting the higher impact of cement, water and curing time parameters in all two types of strength. In terms of seashell content, the results confirmed that adding high amounts of the seashell to the concrete decreases its strength, while small amounts can be used without any significant adverse effect on strength parameters.

Graph neural networks for parameter estimation in micro-electro-mechanical system testing

Micro-electro-mechanical systems (MEMS) are of great importance in a broad range of applications including vehicle safety and consumer electronics. During the testing of these devices, large heterogeneous data sets containing a variety of parameters are recorded. Aiming to substitute costly measurements as well as to gain insight into the relations among the measured parameters, graph neural networks (GNNs) are investigated. Thus, the questions are addressed whether for inference of MEMS final module level test parameters, working on graph structures leads to an improvement of the predictive performance compared to the analysis via standard machine learning approaches on tabular data and how the graph structure and learning algorithm contribute to the overall performance. To evaluate this, in an empirical study different graph representations of the acquired test data were set up. On these, four different state-of-the-art GNN architectures were trained and compared on the task of raw sensitivity prediction for a MEMS gyroscope. Whereas the GNNs performed on par with a light gradient boosting machine, neural network and multivariate adaptive regression splines model used as baseline on the complete data set, in the presence of sparse data, the GNNs outperformed the baseline methods in terms of the overall root-mean-square error (RMSE) and achieved distinct improvement in the maximum error when trained on data with similar sparsity rates as observed during the validation. • Proposal and demonstration of GNNs for parameter estimation in MEMS testing. • Application to sensitivity estimation at final module testing of MEMS gyroscopes. • Evaluation of different graph construction variants and four GNN architectures. • Distinct performance improvement in the case of sparse measurement data.

Investigation of affective traits affecting mathematics achievement by SEM and MARS methods

The purpose of the study is to analyze the affective traits that affect mathematics achievement through Structural Equation Modeling (SEM) as a traditional regression model and Multivariate Adaptive Regression Splines (MARS), as one of the data mining methods. Structural Equation Modeling, one of the regression-based methods, is quite popular for social sciences due to the various advantages it offers; however, it requires very intensive assumptions. MARS method, on the other hand, is a multivariate and adaptive nonparametric statistical regression method used for data classification and modeling. MARS does not need any assumptions such as normality, linearity, homogeneity. It allows variables that do not provide linearity to be included in the analysis. The present study examines whether it is possible to use the MARS method, which is a more flexible method compared to SEM, taking both methods into account. Regarding this goal, the SEM model was created with the program R using the affective data and the achievement variable picked from TIMMS 2019 data. Then, the MARS method was created using the SPM (Salford Predictive Modeler) program. The results of the study showed that at certain points the MARS model gave similar results to the SEM model and MARS model is more compatible with the literature.

Predicting grain protein concentration in winter wheat (Triticum aestivum L.) based on unpiloted aerial vehicle multispectral optical remote sensing

ABSTRACT Prediction models for crude protein concentration (CP) in winter wheat (Triticum aestivum L.) based on multispectral reflectance data from field trials in 2019 and 2020 in southern Sweden were developed and evaluated for independent trial sites. Reflectance data were collected using an unpiloted aerial vehicle (UAV)-borne camera with nine spectral bands having similar specification to nine bands of Sentinel-2 satellite data. Models were tested for application on near-real time Sentinel-2 imagery, on the prospect that CP prediction models can be made available in satellite-based decision support systems (DSS) for precision agriculture. Two different prediction methods were tested: linear regression and multivariate adaptive regression splines (MARS). Linear regression based on the best-performing vegetation index (the chlorophyll index) was found to be approximately as accurate as the best performing MARS model with multiple predictor variables in leave-one-trial-out cross-validation (R2 = 0.71, R2 = 0.70 and mean absolute error 0.64%, 0.60% CP respectively). Models applied on satellite data explained to a small degree between-field variations in CP (R2 = 0.36), however did not reproduce within-field variation accurately. The results of the different methods presented here show the differences between methods used and their potential for application in a DSS.

Application of Multivariate Adaptive Regression Splines (MARS) approach in prediction of compressive strength of eco-friendly concrete

Concrete is the most often used material in the building sector. The use of ground-granulated blast-furnace slag (GGBFS) and recycled concrete aggregates (RCA) in concrete reduces its negative environmental impact. However, increasing the number of ingredients in concrete complicates the prediction of its compressive strength. This research aims to develop a machine learning model for predicting the compressive strength of eco-friendly concrete using Multivariate Adaptive Regression Splines (MARS). MARS is a well-known technique for creating predictive modeling equations using experimental data. A data set of 161 concrete specimens was obtained to train and evaluate the machine learning algorithms for this work. The findings of the MARS prediction algorithm were compared to those of the Support Vector Machine and the Random Forest regression algorithms (black-box models) using the training set. A hyperparameter tuning procedure was applied to find the optimal MARS, RF, and SVM model parameters. The results of 5-fold cross-validation indicate that the MARS optimal model has the highest coefficient of determination (R2), with a value equal to 0.889, and the lowest Root Mean Squared Error (RMSE) measure, with a value equal to 4.110 MPa. For our case, the performance metrics measures proved the superiority of the MARS model when compared with RF and SVM. The final MARS optimal prediction equations were built using a full training set and were then evaluated using unseen data (testing set). Our model has the potential to help civil engineers in designing durable infrastructures.

Quantitative driving analysis of climate on potential evapotranspiration in Loess Plateau incorporating synergistic effects

Potential evapotranspiration (ET0) is an important part of regional hydrological cycle and energy cycle. In the context of global climate change, it is of great significance to reveal the spatiotemporal variation of ET0 and its response to climate change for further understanding the evolution mechanism of water cycle in changing environment. In this study, the Penman-monteith formula was utilized to calculate ET0 in 6 sub-basins of the Loess Plateau (LP) during 1974–2019 using the daily observed data from 84 meteorological stations. The Mann-Kendall trend test and Kriging interpolation were applied to analyze the temporal and spatial characteristics of ET0 on annual and seasonal scales. In addition, the multivariate adaptive regression splines model, sensitivity analysis and contribution rate were used to identify the driving factors of ET0 variations by considering the synergistic effect of multiple factors. Results showed that (1) The ET0 of LP showed an insignificant increase at a rate of 5.448 mm/10a on the annual scale. At the seasonal scale, ET0 increased in spring and winter, but decreased in summer and autumn. Spatially, the annual average, spring, summer and autumn ET0 showed a trend of higher north–south and lower east–west; (2) The annual ET0 of LP is most sensitive to relative humidity (RH). At the seasonal scale, ET0 was most sensitive to sunshine hours(S) in spring and autumn, average temperature (T) in summer and RH in winter; (3) RH is the main influencing factor of ET0 change in the LP. At the seasonal scale, wind speed (WS) and T dominated the ET0 variation in spring and winter, respectively. S dominated the ET0 variation in summer and autumn; (4) The variation trend and driving factors of ET0 in each sub-catchment were different from those in the LP, and there were significant differences among the sub-catchments. This study provides a reference for the sustainable development of vegetation restoration and the management of agricultural water resources in the LP.

ANALISIS SURVIVAL PENDERITA GAGAL GINJAL DENGAN PENDEKATAN MULTIVARIATE ADAPTIVE REGRESSION SPLINE

Survival analysis is a statistical method where the outcome variable that is considered is the time until the occurrence of an event or called an event. Survival analysis in the medical world can be used to test the survival of patients suffering from a disease, one of which is kidney failure. Based on the results of the 2013 Basic Health Research, 0.2% of the total population of Indonesia, namely 499,800 people, experienced kidney failure. Kidney failure is a process of decreasing kidney function that requires renal replacement therapy, in the form of hemodialysis or kidney transplantation. If the survival data involve independent variables that are thought to affect the survival time, then the Cox Proportional Hazard (PH) regression approach can be developed which can be developed for the Multivariate Adaptive Regression Spline (MARS) approach. The MARS approach is a nonparametric regression approach with high-dimensional data, namely data with an independent variable of 3≤𝑝≤20 and a data sample of 50≤N≤1000. This study aims to describe the characteristics of patients with kidney failure, identify the factors that affect the survival time of patients with kidney failure and determine the MARS model. The MARS approach is obtained from a combination of functional basis, maximum interaction, and minimum observation by trial and error. Based on the model that has been formed using the MARS approach, it is found that the most influential variables on the survival time of patients with kidney failure are age, urea levels, and creatine levels.&#x0D; Key words: Renal failure, hemodialysis, survival analysis, nonparametric, MARS&#x0D;

Predicting the Mortality of ICU Patients by Topic Model with Machine-Learning Techniques

Predicting clinical patients’ vital signs is a leading critical issue in intensive care units (ICUs) related studies. Early prediction of the mortality of ICU patients can reduce the overall mortality and cost of complication treatment. Some studies have predicted mortality based on electronic health record (EHR) data by using machine learning models. However, the semi-structured data (i.e., patients’ diagnosis data and inspection reports) is rarely used in these models. This study utilized data from the Medical Information Mart for Intensive Care III. We used a Latent Dirichlet Allocation (LDA) model to classify text in the semi-structured data of some particular topics and established and compared the classification and regression trees (CART), logistic regression (LR), multivariate adaptive regression splines (MARS), random forest (RF), and gradient boosting (GB). A total of 46,520 ICU Patients were included, with 11.5% mortality in the Medical Information Mart for Intensive Care III group. Our results revealed that the semi-structured data (diagnosis data and inspection reports) of ICU patients contain useful information that can assist clinical doctors in making critical clinical decisions. In addition, in our comparison of five machine learning models (CART, LR, MARS, RF, and GB), the GB model showed the best performance with the highest area under the receiver operating characteristic curve (AUROC) (0.9280), specificity (93.16%), and sensitivity (83.25%). The RF, LR, and MARS models showed better performance (AUROC are 0.9096, 0.8987, and 0.8935, respectively) than the CART (0.8511). The GB model showed better performance than other machine learning models (CART, LR, MARS, and RF) in predicting the mortality of patients in the intensive care unit. The analysis results could be used to develop a clinically useful decision support system.

Quantifying uncertainty of subsampling-based ensemble methods under a U-statistic framework

ABSTRACT This paper addresses the problem of variance estimation of predictions obtained from a subsampling-based ensemble estimator, such as subbagging and sub-random forest. We first recognize that a subsampling-based ensemble can be written as an infinite-order U-statistic of degree , where is the subsample size that may depend on the learning sample size n. As a result, one can study the uncertainty of predictions obtained from a subsampling-based ensemble under a U-statistic framework, such as approximating its asymptotic variance. However, existing methods used to estimate the asymptotic variance relies on some regularity conditions. In addition, they tend to yield variance estimations with large bias in finite sample scenarios. Motivated by the work of Wang and Lindsay (2014), we propose to construct an unbiased variance estimator for a subsampling-based ensemble. It is efficient to realize with the help of a partition-resampling scheme. We show by simulation studies that the proposed variance estimator yields better performance in terms of mean, standard deviation, and mean squared error compared to both the infinitesimal jackknife and internal variance estimation methods under either a simple linear regression model or a multivariate adaptive regression splines model. Furthermore, we present how to construct an asymptotic confidence interval for the expected prediction at a given test instance using the proposed variance estimator, and compare its coverage probability to that of competing methods. In the end, we demonstrate the practical applications of the methodology using two real data examples.

Modeling the flow rate of dry part in the wet gas mixture using decision tree/kernel/non-parametric regression-based soft-computing techniques

Owing to its importance in extraction of natural gas from underground gas storage as well as its crucial role in determination of final gas mixture in the production facilities of gas/oil industry, the dry content of wet gas mixture needs to be calculated precisely. The present study explores the potential of different soft-computing techniques in estimation of the dry gas flow rate (kg/h) (output variable) of wet gas mixture based on two input variables of wet gas flow rate (kg/h) and absolute gas humidity (g/m3). Decision tree-based methods (M5P tree, random forest (RF), random tree (RT), and reduced error pruning tree (REPT) models), kernel function-based approaches (Gaussian process regression (GPR) and support vector machines (SVM)), and non-parametric regression-based technique (multivariate adaptive regression splines (MARS)) were implemented for the first time to estimate the dry gas flow rate, and their respective prediction performances were analyzed statistically. Coefficient of correlation (CC), Nash–Sutcliffe efficiency (NSE), root mean square error (RMSE), mean absolute error (MAE), Legates and McCabe's index (LMI), and Willmott's Index (WI) were used as the statistical indicators for validating the performance of each soft-computing model. While M5P model (MAE = 122.2382 kg/h, RMSE = 580.5626 kg/h, CC = 0.9875 for the testing data set) was better than other tree-based models (MAE = 363.2802–542.6119 kg/h, RMSE = 871.9363–1025.3444 kg/h, CC = 0.9587–0.9706 for the testing data set) and MARS model (MAE = 128.0083 kg/h, RMSE = 622.9515 kg/h, CC = 0.9852 for the testing data set), the statistical indicators approved the superiority of the radial basis kernel function-based GPR model (GPR-RBKF) model (MAE = 163.3266 kg/h, RMSE = 483.1359 kg/h, CC = 0.9915 for the testing data set) over other implemented models in predicting the dry gas flow rate. The findings highlighted the potential of soft-computing methodologies in precise estimation of dry gas flow rate in wet gas mixture, particularly, in situations where the measurement of such parameters with traditional deterministic models is practically not possible.

On the implementation of the interpretable data-intelligence model for designing service life of structural concrete in a marine environment

Modeling the durability design of reinforced concrete structures is a particularly crucial challenge for identifying the behavior of marine concrete. In this research, novel interpretable equation-based models based on Fick's second law of diffusion for designing the service life of structural concrete were presented. To do so, a Machine Learning (ML) approach based on linear and non-linear methods including Multivariate Adaptive Regression Splines (MARS), Gene Expression Programming (GEP), and M5p Model Tree (MT) was performed to predict the apparent surface chloride concentration (Cs) in the zones of the marine environment. A comprehensive database is comprising of 642 field exposure experimental and environmental records was utilized to develop ML methods. The proposed ML models were investigated in terms of performance evaluation, error measurements, and visual consideration, MARS model (r=0.890,WI=93.9%andRMSE=0.804%bindercontent, (tidal zone) and r=0.901,WI=93.5%andRMSE=0.818%bindercontent (splash zone) and r=0.875,WI=86.8%andRMSE=0.883%bindercontent (submerged zone)) outperformed the proposed GEP, MT and empirical equations. Furthermore, simulation of Monte-Carlo analysis was extended to verify the proposed ML predictive models. Evaluation of the sensitivity analysis presented that the water to binder ratio, annual mean temperature, and exposure time were the most influential factors in predicting the Cs of reinforced concrete structures. Also, a parametric assessment has been performed for the presentation of the robustness of the proposed ML model. The modeling results presented new insight into the designing of the service life of reinforced concrete with superiority promotion of ML methods.

Wheat Yield Prediction in India Using Principal Component Analysis-Multivariate Adaptive Regression Splines (PCA-MARS)

Crop yield forecasting is becoming more essential in the current scenario when food security must be assured, despite the problems posed by an increasingly globalized community and other environmental challenges such as climate change and natural disasters. Several factors influence crop yield prediction, which has complex non-linear relationships. Hence, to study these relationships, machine learning methodologies have been increasingly adopted from conventional statistical methods. With wheat being a primary and staple food crop in the Indian community, ensuring the country’s food security is crucial. In this paper, we study the prediction of wheat yield for India overall and the top wheat-producing states with a comparison. To accomplish this, we use Multivariate Adaptive Regression Splines (MARS) after extracting the main features by Principal Component Analysis (PCA) considering the parameters such as area under cultivation and production for the years 1962–2018. The performance is evaluated by error analyses such as RMSE, MAE, and R2. The best-fitted MARS model is chosen using cross-validation and user-defined parameter optimization. We find that the MARS model is well suited to India as a whole and other top wheat-producing states. A comparative result is obtained on yield prediction between India overall and other states, wherein the state of Rajasthan has a better model than other major wheat-producing states. This research will emphasize the importance of improved government decision-making as well as increased knowledge and robust forecasting among Indian farmers in various states.

Evaluation of Logistic Regression and Multivariate Adaptive Regression Spline Models for the Analysis of Factors Determining Entrepreneurial Success

The present research work aims at evaluating Logistic Regression (LR) and Multivariate Adaptive Regression Spline (MARS) models through comparing the results of analysis of factors contributing to entrepreneurial success in Madurai District. The study considered two financial indicators namely Profit Rate and Reinvestment Rate for defining success in entrepreneurial venture. It analysed five factors influencing entrepreneurial success among micro and small scale manufacturing enterprises in the district. Previous business experience was the most important factor contributing to entrepreneurial success, followed by Entrepreneur’s Training and Entrepreneurial orientation. Using Number of significant factors, Accuracy, Sensitivity and Specificity, the results of LR and MR models have been compared. The study validates that MARS model is more powerful and reliable than LR model in predicting the possibility of entrepreneurial success in the business venture.