Abstract: The construction industry is a very important sector of the Nigerian economy. It contributes significantly to the Gross National Product. Cost overrun is an integral part of the construction industry. It generates in projects financial loss to both contractors and owners (clients). It is an important parameter for success of any project that results to serious sequences. Cost overrun is a chronic problem for tertiary institutions. This is because, it is common for projects not to be completed on time and within the initial project budget. The paper assess the management of project cost overrun, reasons for cost overrun and suggested solutions in selected Public Tertiary Institutions in Lafia Metropolis using PartialLeastSquares-StructuralEquationModelling (PLS-SEM). The results show that contractor’s site management related factors has 97.6% effect on cost overrun, followed by non-human resource related factors with an effect of 94.4% on cost overrun. The least was information and communication technology related factors having 75.7% effect on cost overrun. The findings reveal that the PLS-SEM is a model that evaluates a data as a collective entity.
Abstract: Most of the countries globally depend on the energy-related industries in daily and economic activities. Due to the perpetual population growth and the fast economic development, the development of energy generation model is crucial in availing the economic planning strategy of the country. The study developed and evaluated the energy generation model of Al-Zawiya Steam Power Plant, Libya. Specifically, the study focused on the causal relationship between input and output of the energy generation model using the PartialLeastSquares–StructuralEquation Model (PLS-SEM) method as the sample size was too small to utilize StructuralEquationModelling-Analysis of Moment Structure (SEM-AMOS). A data was gathered from Al-Zawiya Steam Power Plant, Libya which consisted of 12 indicator variables with 60 observations. The analysis revealed that most of the causal relationships in the developed model were significant at p<0.005. The results indicated that the developed model was strengthened by empirical analysis and in parallel with the preceding findings and theoretical framework. Apart of input and output structural model, the study also prosperously validated all the indicator variables depicted in input and output measurement model. In conclusion, this study had successfully developed and evaluated energy generation model and corroborated the causal relationship of several input and output latent variables by betokens of structuralequation model through PLS-SEM approach.
The results for the capital structure choice through the mediation effects of leverage are discussed in Figure 1. This mediation effects can be concluded to be either “none,” “partial,” or “full” mediation of the three path coefficient estimates. “None mediation” effect is when there is a non-significant value for all path estimators. “Partial mediation” is when the path estimates for direct effects are all significant as well as indirect significant. “Full mediation” is when the indirect effect is significant but the direct effects (c’) are no sign of their significant value (Baron and Kenny, 1986; Iacobucci and Duhachek, 2003; MacKinnon et al., 1995). This study concludes that both factors of the capital structure choice have partial mediating effects because they meet the adaption from the three conditions proposed by Baron and Kenny (1982) except for the IR that shows a full mediation effects. For example, the firms with high tangibility assets have higher tendency to face the financial distress since the assets (such as property, plant and equipment) are involved in the process of a productive resource due to the tendency to attain a high liquidation value. This relationship is also supported by the trade-off theory (TOT) and the POT and AT
Abstract: Institution of higher learning in Malaysia has implemented an Outcome-Based Education (OBE) since 2007. Among the crucial components of OBE were the Program Education Objectives (PEOs) and Program Learning Outcomes (PLOs). The development of PEOs has to take into consideration the involvement of stakeholders, in order to address their needs and requirements and also must in line with the institutional vision and mission. PLOs, on the other hand, has to be aligned with PEOs to attain OBE. The study withal strive the relationship between PLOs and PEOs in OBE by using PartialLeastSquares-StructuralEquation Model (PLS-SEM) approach as the sample size was too small to utilize StructuralEquationModelling-Analysis of Moment Structure (SEM-AMOS). A simple random sampling was used to select 90 teachers throughout Malaysia. The selected teachers were a graduate of Sultan Idris Education University's (UPSI) Mathematics Education Degree (BEd Maths). From the analysis, it revealed that all the relationships in the developed model were significant at p<0.001. The results indicated that the developed model was strengthened by empirical analysis and in parallel with the preceding findings and theoretical framework. Apart of PEOs and PLOs path model, the study also successfully validated all the indicator variables depicted in PEOs and PLOs measurement model. This study had successfully developed and evaluated PEOs and PLOs relationship in OBE by means of structuralequation model through PLS-SEM approach. In conclusions, the relationship between PEOs and PLOs can not only be expressed qualitatively, but also be modeled in the structuralequation.
In addition to relationships between variables being incorporated into structuralequation models it is also possible to incorporate means (or, in the case of endog- enous variables, intercepts). In the RAM specification, a mean is modelled as a triangle. The path diagram shown in Figure 4 effectively says "estimate the mean of the vari- able x". In the circumstances where there are no restrictions, the estimated mean of x in the model will be the mean of x in the data. However, it is possible to place restrictions on the data, and test the model, again with a χ 2 test. A simple example would be to place a restriction
Classical PCR, PLS and RR techniques are well known shrinkage estimators designed to deal with multicollinearity (see, e.g., Frank and Friedman, 1993, Montgomery and Peck, 1992, Jolliﬀe, 1986). The multicollinearity or near-linear dependence of regressors is a serious problem which can dramatically inﬂuence the eﬀectiveness of a regression model. Multicollinearity results in large variances and covariances for the leastsquares estimators of the regression coeﬃcients. Multicollinearity can also produce estimates of the regression coeﬃcients that are too large in absolute value. Thus the values and signs of estimated regression coeﬃcients may change considerably given diﬀerent data samples. This eﬀect can lead to a regression model which ﬁts the training data reasonably well, but generalizes poorly to new data (Montgomery and Peck, 1992). This fact is in a very close relation to the argument stressed in (Smola et al., 1998), where the authors have shown that choosing the ﬂattest linear regression function 1 in a feature space can, based on the smoothing properties of the selected kernel function, lead to a smooth nonlinear function in the input space.
Feature extraction techniques are methods widely used to reduce the dimensionality of a data while retaining most of the relevant information in the original data. Locality preserving partialleastsquares (LPPLS) is a recently developed feature extraction technique that aims to preserve the local structural information of data. LPPLS seeks to preserve local structure defined by nearest neighbors. However, the nearest neighbors may belong to different classes which might lead to the poor performance of LPPLS in discriminating the different classes in the data. In this paper, we propose an extension of LPPLS called extended locality preserving partialleastsquares which consider class label information. The binary (0-1) weighting technique together with label information is used to construct the similarity matrices that determine local projection of the data. Therefore, our extended LPPLS does not simply preserve local structure, but also has discriminating power to differentiate data from different classes. Experimental results on various data sets demonstrate the effectiveness of the proposed extended LPPLS. Two different evaluation metrics, normalized mutual information (NMI) and Fowlkes-Mallow index are used to measure the accuracy of methods used in the experiments. Keywords: Class labels, feature extraction, local information, similarity matrix.
An alternative to PLS-SEM was introduced by   as generalized struc- tured component analysis (GSCA). We apply GSCA as a robustness test because it belongs to the same family of methods. Both PLS-SEM and GSCA are va- riance-based methods appropriate for predictive modelling and they substitute components for factors. GSCA retains the advantages of PLS-SEM such as fewer restrictions on distributional assumptions ( i.e. multivariate normality of ob- served variables is not required for parameter estimation), unique component score estimates, and avoidance of improper solutions with small samples , . As reference (, p.174) clearly point out “…comparison of PLS to other methods cannot and should not be applied indiscriminately.” We re-state that CB-SEM is not a feasible or meaningful alternative to PLS-SEM under the con- ditions of the current study, where the sample size is small compared to the population, formative indicators are present and the theorised model is explora- tory.
The quality of modelling can be compared with that of other recent, relevant studies. Lomborg et al.  used a wide range of approximately 100 whole straw samples (down-sampled, milled to 1 mm) from a variety of sources and different seasons to explore the use of FT-NIR spectroscopy in determining chemical composition. They reported %RMSEP (root mean square of prediction) values of 11% for glucan and xylan, 13% for arabinan and 12% for lignin, using 5, 5, 4 and 7 PLS factors respectively. This relied on heavy use of outlier rejection (as much as 18% for lignin). A subsequent FT-NIR rapid analysis study by Liu et al.  on corn and switchgrass (not wheat straw) gave lower relative errors of 1.99, 2.3, 10.96, 7.53, 6.65, 3.62 and 13.95% for glucan, xylan, galactan, arabinan, mannan, lignin and ash. FTIR has an advan- tage over FT-NIR in that much more chemical informa- tion is shown by the fundamental vibrations.
StructuralEquation Modeling (SEM) with latent variables is becoming increasingly popular in social and behavioral science (Boomsma, 2000) . The literature on SEM distinguishes between two different operationalizations of the relationships between latent variables and their observed indicators: the reflective (principal factor) and the formative (composite index) measurement models of latent variable. Numerous studies have by default or erroneously by design incorrectly specified their items as reflective when they should have used a formative measurement model operationalization (Jarvis, MacKenzie, & Podsakoff, 2003). This is somewhat surprising considering the fact that the understanding of formative indicator orientation is not new (Blalock, 1971) and previous research has focused on the nature, identification, and validation issues of formative indicators (Bollen & Lennox, 1991; Diamantopoulos & Winklhofer, 2001; Edwards & Bagozzi, 2000; MacCallum & Browne, 1993).
relationship between the Vis/SWNIR spectra and the bruise susceptibility of apple need to be developed. To improve the generalization ability of the model, this work used the selective ensemble learning model based on feature selection (SELFS) to predict bruise susceptibility in apples. The SELFS mainly consists of individual generation, individual learner, and conclusion synthesis (denoted as I, II, and III in Figure 6). The individual generation is mainly used to generate a certain number of calibration subsets from whole calibration samples for developing individual learner. In this study, the calibration subsets were produced using feature selection method based on successful projection algorithm (SPA). The individual learner is mainly designed to establish a sub-model for each calibration subset. Partialleast square (PLS) regression is a widely used chemometric method for building prediction model. PLS is suited well when the matrix of predictors has more variables than observations, and also when the input variables contain noise and are strongly correlated  . Therefore,
Evaluation index. This study established cross- validation models by NIRS that is separately combined with PLS, iPLS, siPLS, and biPLS. With RMSECV as the main standard, the correlation coefficient (r) and bias as reference, the best modelling parameters in each PLS model and the most ideal one overall were selected. The spectral range of the PLS model was 400 nm to 2500 nm, with RMSECV, RMSECP values of 0.0637, 0.0630 for the light soy sauce samples and 0.1043, 0.1040 for the dark soy sauce samples, re- spectively. RMSECV(P) is principally used to assess the feasibility of modelling methods and the predic- tion ability of the corresponding models to present accurate results with low RMSECV(P).
With the exponentially growing volume of data sets, multivariate methods for reducing dimensionality are an important research area in statistics. For combining two data sets, PartialLeastSquares (PLS) regression  is a popular dimension reduction method . PLS decomposes variation in each data set in a joint part and a residual part. The joint part is a linear projection of one data set on the other that best explains the covariance between the two data sets. These projections are obtained by iterative algorithms, such as NIPALS . PartialLeastSquares is popular in chemometrics . In this field, the focus is on development of algorithms with good prediction performance, while the underlying model is less important. For applications in life sciences, interpretation of parameter estimates is necessary to gain understanding of the underlying molecular mechanisms.
A new quantitative method by using the data obtained from chemometrics calibration partialleastsquares (PLS), principal component regression (PCR), and inverse leastsquares (ILS) was developed for the simultaneous analysis of ET and TC without any separation process in combined pharmaceutical products. A Thermo Scientific Multiscan GO 51119300 model UV-Visible spectrophotometer was used for spectrophotometric measurements. The spectral bandwidth was settled at 1 nm, and the wavelength scan- ning speed of the proposed method was 1000 nm/min. PLS, PCR, and ILS algorithms were conducted by using MATLAB R2014b (PLS-Toolbox software version 8.1.1). Design-Expert 8.0 (Stat-Ease Inc., Minneapolis, MN, USA)
In this work we describe the leastsquares matrix (LSM) algorithm for the estimation of the dynamic state-space models. In conventional recursive leastsquares (RLS) algorithms uptake is based on the use of the forgetting factor, which weights the data vectors by an exponent- tially descending function. In the present algorithm the rectangular weighting function is used, where the Hankel data matrix includes M vectors (4). The uptake of the data matrix consists of the addition of the most recent data vector n 1 and subtraction of the latest vector 1 . This leads to a novel uptake mechanism (7) via
have a J-shape with a minimum COT within the range of the ordinary U (Claireaux et al., 2006). The J-shaped curve of COT in fish is possibly derived from the highly non-linear relationship of V O2 vs U, which is represented by the exponent c in Eqn 4 significantly greater than the unity. The minimum COT corresponds to the optimum efficiency of locomotion (Videler, 1993). If we take non-unity c (1.67) into the calculation of COT in Paramecium swimming obtained from the least-squares fitting of Eqn 4, the minimum COT would be found at 1.45 mm s –1 , which is nearly the maximum speed. It is, therefore, argued that the efficiency of energy expenditure of swimming in Paramecium increases with speed. This is the second characteristic of Paramecium swimming, which we discovered for the first time from experiments carried out in this study, which have enabled us to measure oxygen consumption and swimming speed simultaneously.
Data analysis techniques make use of in this research were StructuralEquation Modeling (SEM) using the PartialLeast Square (PLS) method. The research instrument was used to compile the questionnaire. The questionnaire that was used in this research was Likert scale. The sampling technique obtained 150 people. The samples chosen were the sub-districts that has the most female population in Karanganyar Regency. The research sample consisted of the people of Matesih District, Jumantono District, and Karanganyar District.
Finally, this study is produce a structural model showing how these factors interfere each other so it may cause the obesity in adolescent. From the model it can be determined the relationship o f each factors with the weight o f adolescent. In addition, it can also determine the relationship among the factors itself. Moreover, as time goes by, the learning process is already exploiting these technological advances to learn and know more on obesity. This research expects to produce an e-learning application based on the result from the statistical analysis with the PLS-SEM. This is to produce an obesity rehabilitation learning application accordance with the existing situation.
The partialleast square StructuralEquation modeling is carried out in SMART PLS 3 software. The formulated structuralequation model is given in the figure 1. Bootstrapping with complete bootstrapping option is carried out to find standard error and the T-values. Comparing the T-Value with 95% percentage confidence value of 1.96, the above hypotheses are tested.