Multivariate adaptive regression splines and neural network models for prediction of pile drivability

Zhang, Wengang; Goh, Alvina

doi:10.1016/j.gsf.2014.10.003

Cited by 423 publications

(149 citation statements)

References 12 publications

Supporting

Mentioning

141

Contrasting

Unclassified

Order By: Relevance

“…The MARS method is one of the numerous data-mining tools used for solving regression problems [13][14][15][16]. The method is an extension of the classic approach to input data in the regression model.…”

Section: Methodsmentioning

confidence: 99%

Assessment of the Effect of Wastewater Quantity and Quality, and Sludge Parameters on Predictive Abilities of Non-Linear Models for Activated Sludge Settleability Predictions

Szeląg

Gawdzik

2017

Pol. J. Environ. Stud.

View full text Add to dashboard Cite

Section: Methodsmentioning

confidence: 99%

Assessment of the Effect of Wastewater Quantity and Quality, and Sludge Parameters on Predictive Abilities of Non-Linear Models for Activated Sludge Settleability Predictions

Szeląg

Gawdzik

2017

Pol. J. Environ. Stud.

View full text Add to dashboard Cite

“…The strengths of the CHAID method include (1) the Chi-Square method used for attribute, (2) nominal types and interval variables can be considered as predictors, (3) continuous variables can be chosen as criterion variables and (4) a criterion variable can be established [53].  MARS: Multivariate Adaptive Regression Splines (MARS) is a flexible procedure to organize relationships between a set of input variables and the target dependent variables [54]. The MARS algorithm is non-linear and non-parametric regression method [55] is based on the divide and conquer strategy where training data sets are "partitioned into separate piecewise linear segments (splines) of differing gradients (slope) " [54].…”

Section: B Decision Treesmentioning

confidence: 99%

“… MARS: Multivariate Adaptive Regression Splines (MARS) is a flexible procedure to organize relationships between a set of input variables and the target dependent variables [54]. The MARS algorithm is non-linear and non-parametric regression method [55] is based on the divide and conquer strategy where training data sets are "partitioned into separate piecewise linear segments (splines) of differing gradients (slope) " [54]. The MARS algorithm "divides the space of predictors into multiple knots and then fits a spline function between these knots" [56].…”

Section: B Decision Treesmentioning

confidence: 99%

Design and Development of an optimal algorithm to assign applicants to suitable teaching positions

Chishimba

Kunda

2018

zictjournal

View full text Add to dashboard Cite

Resource allocation has always been an area of interest in the area of computing. Areas such as machine learning provide many solutions to the problem of resource allocation. The issue addressed in this study is the issue of optimal allocation of applicants (teachers) to positions in schools where their area of specialization will be better applied. We develop an algorithm that is able to allocate applicants to schools based on the applicant qualifications and the school's needs. We use the principles of resource allocation and machine learning in order to create an application to allocate applicants to schools where their qualifications are most suited. Methods used include classification techniques in machine learning, regression and similarity comparison. For the identification of subjects an applicant in proficient in, various machine learning algorithms are tested to determine which machine learning algorithm was best. The actual process of identifying which applicant qualifies for a school position is also tested against sequential assignment if applicants to schools. The results of this was the algorithm based assignment of applicants to schools which produced more accurate assignment of applicants to schools than the sequential assignment of applicants. The aim of this algorithm is to provide a solution to that automatically identifies the needs (subjects) of a school, determine which needs have a higher priority, identify the qualifications of the applicants and assign the applicants to the school according to the school's needs and the applicant's qualifications.

show abstract

“…Dynamical behaviours for the neural networks especially for the memristive neural networks have attracted ascending attention recently, [5][6][7][8][9][10][11][12][13][14][15][16] among which, the synchronization control of fractional-order memristive neural networks was considered in Bao et al, 5 an interval matrix method was employed in Yang et al, 7 based on this new method, the memristive system can be equivalent to written as a system with uncertain systems, via a saturating sampled-data controller, the stabilization control of memristive neural networks was considered in Ding et al, 8 and Wu and Zeng 16 pays attention to the passivity analysis of memristive neural networks. The passivity theory means that the systems cannot consume more energy than what they absorb, ie, it can keep a system internally stable, and therefore plays a key role in stability analysis of nonlinear dynamical systems.…”

Section: Introductionmentioning

confidence: 99%

Passivity and passification of quaternion‐valued memristive neural networks

Wei

2019

Math Methods in App Sciences

View full text Add to dashboard Cite

In this paper, a class of memristive neural networks with quaternion-valued connection weights is studied. By starting from some basic quaternion-valued algorithms, the quaternion-valued memristive system is obtained; then, some passivity criteria for the memristive neural networks are presented based on some appropriate auxiliary functions. Furthermore, to tackle with the passification problem, two kinds of control protocols are designed. What should be mentioned is that, in the above derived conclusions, the partial order is employed, which can be employed to compare the "magnitude" of two different quaternions, and thus, the closed convex hull consisted by quaternion can be derived correspondingly. In the end, the analytical design are substantiated with two numerical results.

show abstract

Multivariate adaptive regression splines and neural network models for prediction of pile drivability

Cited by 423 publications

References 12 publications

Assessment of the Effect of Wastewater Quantity and Quality, and Sludge Parameters on Predictive Abilities of Non-Linear Models for Activated Sludge Settleability Predictions

Assessment of the Effect of Wastewater Quantity and Quality, and Sludge Parameters on Predictive Abilities of Non-Linear Models for Activated Sludge Settleability Predictions

Design and Development of an optimal algorithm to assign applicants to suitable teaching positions

Passivity and passification of quaternion‐valued memristive neural networks

Contact Info

Product

Resources

About