Considering the complexities and challenges in the classification of multiclass and imbalanced fault conditions, this study explores the systematic combination of unsupervised and supervised learning by hybridising clustering (CLUST) and optimised multi-layer perceptron neural network with grey wolf algorithm (GWO-MLP). The hybrid technique was meticulously examined on a historical hydraulic system dataset by first, extracting and selecting the most significant statistical time-domain features. The selected features were then grouped into distinct clusters allowing for reduced computational complexity through a comparative study of four different and frequently used categories of unsupervised clustering algorithms in fault classification. The Synthetic Minority Over Sampling Technique (SMOTE) was then employed to balance the classes of the training samples from the various clusters which then served as inputs for training the supervised GWO-MLP. To validate the proposed hybrid technique (CLUST-SMOTE-GWO-MLP), it was compared with its distinct modifications (variants). The superiority of CLUST-SMOTE-GWO-MLP is demonstrated by outperforming all the distinct modifications in terms of test accuracy and seven other statistical performance evaluation metrics (error rate, sensitivity, specificity, precision, F score, Mathews Correlation Coefficient and geometric mean). The overall analysis indicates that the proposed CLUST-SMOTE-GWO-MLP is efficient and can be used to classify multiclass and imbalanced fault conditions.Article Highlights: The issue of multiclass and imbalanced class outputs is addressed for improving predictive maintenance. A multiclass fault classifier based on clustering and optimised multi-layer perceptron with grey wolf is proposed. The robustness and feasibility of the proposed technique is validated on a complex hydraulic system dataset.

Purpose: Being able to predict the likelihood of a project to overrun its cost before the contract signing phase is crucial in developing the required mitigating measures to avert it. Known parameters that permit the timely prediction of cost overrun provide the basis for such predictions. Therefore, the purpose of this paper is to develop a model for forecasting cost overruns.Design/methodology/approach: Ten predictive variables known before the contract signing phase of a project are identified. Based on a survey approach, information on 321 educational projects completed are compiled. A multiple linear regression analysis is adopted for the model development.Findings: Five variables – initial contract sum, gross floor area, number of storeys, source of funds and contractors’ financial classification are observed to influence cost overruns. The model, however, yields a fairly weak coefficient of determination with a mean absolute percentage error of 30.22 and 138 per cent, respectively.Research limitations/implications: The model developed focussed on data only educational projects sampled from three out of the ten administration regions in Ghana based on a purposive sampling approach.Practical implications: Policy makers and construction managers working on public projects stand to gain tremendous assistance in formulating and strengthening their own in-house cost forecasting at the precontract phase based on “what if” analysis to generate various alternative predictions of cost overruns.Originality/value: Considering the innate nature of cost overruns within the Ghanaian construction industry often resulting to project abandonment, this research presents a unique dimension for tackling cost overruns based on a predictive approach.

Knowledge of the dependence between random variables is necessary in the area of risk assessment and evaluation. Some of the existing Archimedean copulas, namely the Clayton and the Gumbel copulas, allow for higher correlations on the extreme left and right, respectively. In this study, we use the idea of convex combinations to build a hybrid Clayton–Gumbel–Frank copula that provides all dependence scenarios from existing Archimedean copulas. The corresponding density and conditional distribution functions of the derived models for two random variables, as well as an estimator for the proportion parameter associated with the proposed model, are also derived. The results show that the proposed model is able to show any case of dependence by providing coefficients for the upper tail and lower tail dependence.

In this extant paper, a multivariate time series model using the seemingly unrelated times series equation (SUTSE) framework is proposed to forecast the peak and short-term electricity demand using time series data from February 2, 2014, to August 2, 2018. Further the Markov Chain Monte Carlo (MCMC) method, Gibbs Sampler, together with the Kalman Filter were applied to the SUTSE model to simulate the variances to predict the next day's peak and electricity demand. Relying on the study results, the running ergodic mean showed the convergence of the MCMC process. Before forecasting the peak and short-term electricity demand, a week's prediction from the 28th to the 2nd of August of 2018 was analyzed and it found that there is a possible decrease in the daily energy over time. Further, the forecast for the next day (August 3, 2018) was about 2187 MW and 44090 MWh for the peak and electricity demands respectively. Finally, the robustness of the SUTSE model was assessed in comparison to the SUTSE model without MCMC. Evidently, SUTSE with the MCMC method had recorded an accuracy of about 96% and 95.8% for Peak demand and daily energy respectively.

Several standard distributions can be used to model lifetime data. Nevertheless, a number of these datasets from diverse fields such as engineering, finance, the environment, biological sciences, and others may not fit the standard distributions. As a result, there is a need to develop new distributions that incorporate a high degree of skewness and kurtosis while improving the degree of goodness-of-fit in empirical distributions. In this study, by applying the T-X method, we proposed a new flexible generated family, the Ramos-Louzada Generator (RL-G) with some relevant statistical properties such as quantile function, raw moments, incomplete moments, measures of inequality, entropy, mean and median deviations, and the reliability parameter. The RL-G family has the ability to model right, left, and symmetric data as well as different shapes of the hazard function. The maximum likelihood estimation (MLE) method has been used to estimate the parameters of the RL-G. The asymptotic performance of the MLE is assessed by simulation analysis. Finally, the flexibility of the RL-G family is demonstrated through the application of three real complete datasets from rainfall, breaking stress of carbon fibers, and survival times of hypertension patients, and it is evident that the RL-Weibull, which is a special case of the RL-G family, outperformed its submodels and other distributions.