Spatiotemporal Patterns in Large-Scale Traffic Speed Prediction

Many intelligent transportation systems (ITS) applications require accurate prediction of traffic parameters. Previous studies have shown that data driven machine learning methods like support vector regression (SVR) can effectively and accurately perform this task. However, these studies focus on highways, or a few road segments. We propose a robust and scalable method using n-SVR to tackle the problem of speed prediction of a large heterogenous road network. The traditional performance measures such as mean absolute percentage error (MAPE) and root mean square error (RMSE) provide little insight into spatial and temporal characteristics of prediction methods for a large network. This inadequacy can be a serious hurdle in effective implementation of prediction models for route guidance, congestion avoidance, dynamic traffic assignment and other ITS applications. We propose unsupervised learning techniques by employing k-means clustering, principal component analysis (PCA), and self organizing maps (SOM) to overcome this insufficiency. We establish the effectiveness of the developed methods by evaluation of spatial and temporal characteristics of prediction performance of the proposed variable window n-SVR method.

Predicting dam inflow is critical for human life safety, water resource management, and hydroelectric power generation. While machine learning (ML) models address complex, nonlinear hydrological problems, quantum machine learning (QML) offers greater potential to overcome classical computational limits. This study compares a hybrid quantum neural network (HQNN) with the following two classical models: bidirectional CNN-LSTM and support vector regression (SVR). These models were evaluated to predict monthly inflow to the Mile Mughan Dam, a transboundary hydroelectric and irrigation dam located on the Aras River between Azerbaijan and Iran, using a 14-year dataset (2010–2023) under two scenarios. In total, 70% of data was used for training and 30% for testing. The first scenario encompassed meteorological variables plus three months of inflow lags, and the second included inflow lags only. Model performance was assessed using Coefficient of Determination (R2), Root Mean Squared Error (RMSE), Nash–Sutcliffe efficiency (NSE), Mean Absolute Percentage Error (MAPE), and graphical plots. HQNN showed superior performance across all metrics. In Scenario 1, HQNN achieved R2 = 0.915, RMSE = 37.318 MCM, NSE = 0.908, MAPE = 8.343%; CNN-BiLSTM had R2 = 0.867, RMSE = 46.506 MCM, NSE = 0.858, MAPE = 10.795%; SVR had R2 = 0.846, RMSE = 52.372 MCM, NSE = 0.821, MAPE = 12.772%. In Scenario 2, HQNN maintained strong performance (R2 = 0.855, RMSE = 48.56 MCM, NSE = 0.845, MAPE = 9.979%) and outperformed CNN-BiLSTM (R2 = 0.810, RMSE = 56.126 MCM, NSE = 0.793, MAPE = 11.456%) and SVR (R2 = 0.801, RMSE = 60.336 MCM, NSE = 0.761, MAPE = 12.901%). In Scenario 1 and Scenario 2, HQNN increased the prediction accuracy by 19.76% and 13.47%, respectively, compared to the CNN-BiLSTM model. These results confirm HQNN’s reliability in both multivariate and univariate modeling.

According to an experimental dataset on the tensile strength and elongation of TA15 titanium alloy under different hot deformation process parameters including temperature, strain, strain rate and cooling condition, support vector regression (SVR) combined with particle swarm optimization (PSO) for its parameter optimization, is proposed to establish a model for prediction of the tensile strength and elongation of hot deformed TA15 titanium alloy. For tensile strength, the mean absolute percentage error (MAPE) achieved by SVR is 0.65% and 0.68% for the training and test set, respectively. At the same time, the MAPE for elongation achieved by SVR is 1.51% and 3.36% for the training and test set, respectively. The MAPEs for both tensile strength and elongation achieved by SVR are much smaller than those of BPNN by using identical training and test samples. Accordingly, the established SVR model was adopted to illustrate the relationships among tensile strength, elongation, and the process parameters. From the 3D surface of tensile strength vs. temperature and strain rate, it is found that to reach a higher tensile strength, a strain rate lower than 0.01s-1 is required, and a lower strain will be helpful for achieving the maximum elongation. These suggest that SVR as a novel approach has a theoretical significance and potential practical value in fabrication of TA15 titanium alloy with desired properties.

Liable, proficient and ecological cognizant electrical energy consumption activities are becoming a basic need for the consistent smart grid. This paper put forwards the concept of a data mining and intelligent proposed model to scrutinize including predict electrical energy time sequences to discover a number of time-based power consumption patterns. Support Vector Regression (SVR) have been productively employed to resolve non-linear regression and time sequences complications associated with prediction of residential electric energy consumption. Jaya algorithm is used in this paper as the implementation of SVR is greatly reliant on the collection of its constraints. The predicting model is technologically advanced by means of weighted SVR configurations (\(\nu \)-SVR and \(\epsilon \)-SVR). Besides, the Jaya algorithm is deployed to decide the weights resultant to every configuration. An instance of time sequential power consumption information from a residential edifice in Denmark is employed to explicate the execution of the presented configuration. Furthermore, the anticipated model is able to estimate power consumption for half hour and daily time successions data for the similar building. The consequences depict that the proposed model demonstrates developed weight for \(\nu \)-SVR for half hour data. Nevertheless, a sophisticated weight for \(\epsilon \)-SVR is perceived for diurnal data. The Mean Absolute Percentage Error (MAPE) for everyday power expenditure data is 5.521 while for half-hour power utilization is 3.769 correspondingly. Also, a thorough evaluation with different algorithms indicate that the presented configuration produces greater exactness for residential power exhaustion prediction.

Support vector regression with genetic algorithms in forecasting tourism demand

Predicting discharge coefficient of triangular labyrinth weir using Support Vector Regression, Support Vector Regression-firefly, Response Surface Methodology and Principal Component Analysis

The primary objective of this study is to evaluate the accuracy of different forecasting models for monthly wind farm electricity production. This study compares the effectiveness of three forecasting models: Autoregressive Integrated Moving Average (ARIMA), Seasonal ARIMA (SARIMA), and Support Vector Regression (SVR). This study utilizes data from two wind farms located in Poland—‘Gizałki’ and ‘Łęki Dukielskie’—to exclude the possibility of biased results due to specific characteristics of a single farm and to allow for a more comprehensive comparison of the effectiveness of both time series analysis methods. Model parameterization was optimized through a grid search based on the Mean Absolute Percentage Error (MAPE). The performance of the best models was evaluated using Mean Bias Error (MBE), MAPE, Mean Absolute Error (MAE), and R2Score. For the Gizałki farm, the ARIMA model outperformed SARIMA and SVR, while for the Łęki Dukielskie farm, SARIMA proved to be the most accurate, highlighting the importance of optimizing seasonal parameters. The SVR method demonstrated the lowest effectiveness for both datasets. The results indicate that the ARIMA and SARIMA models are effective for forecasting wind farm energy production. However, their performance is influenced by the specificity of the data and seasonal patterns. The study provides an in-depth analysis of the results and offers suggestions for future research, such as extending the data to include multidimensional time series. Our findings have practical implications for enhancing the accuracy of wind farm energy forecasts, which can significantly improve operational efficiency and planning.

Accurate prediction of solar irradiation, particularly Diffuse Horizontal Irradiance (DHI), is essential for optimizing solar energy systems. This paper presents a hybrid framework that integrates satellite imagery and satellite-derived data to improve the precision of DHI forecasts. Two key datasets were employed: ground-based Solar Irradiation Measurement (SRM) and satellite-based Solar Irradiation (SSR). The proposed methodology utilizes machine learning models along two paths. The first path processes satellite imagery using advanced techniques, including pixel imputation with a modified Random Forest (RF) and Generative Adversarial Network (GAN), noisy region detection with Self-Organizing Maps (SOM), and noise removal through a Latent Diffusion Model (LDM). The second path handles tabular data through a diffusion probabilistic model designed for missing data imputation. These features are then merged and fed into a Long Short-Term Memory (LSTM) network, enhanced to capture seasonality and trends for accurate DHI predictions. The paper makes significant contributions by introducing a robust hybrid framework that leverages both satellite imagery and tabular data, incorporating novel preprocessing methods such as GAN-based pixel imputation and diffusion-based noise removal. The LSTM model is further adapted to handle seasonal and trend components, resulting in enhanced DHI forecasting accuracy. The proposed model demonstrates superior performance, achieving a Mean Squared Error (MSE) of 8.170 W/m2, Root Mean Squared Error (RMSE) of 1.749 W/m2, Mean Absolute Error (MAE) of 0.693 W/m2, and an R-squared value of 2.574, significantly outperforming conventional methods such as Artificial Neural Networks (ANN), Support Vector Regression (SVR), and traditional LSTM models. Additionally, the model records a low Mean Absolute Percentage Error (MAPE) of 0.8183% and a computation time of 35.4 s, highlighting its efficiency. These results contribute to more accurate and reliable solar energy resource management by offering enhanced DHI forecasts across various spatial and temporal scales.

Forecasting systems reliability based on support vector regression with genetic algorithms

Current and planned solutions used in the ICT infrastructure are largely based on Photonic and RF Communications Systems. Rapid advances in photonic and RF photonic technologies have significantly accelerated the introduction of these new technologies in access storage, backbone and grid networks, defining and building the rules to ensure transfer of multi Tera-bit volumes of data. The purpose of this volume is to provide a vehicle for directing research in this area from multiple lenses in the selected areas of investigation. Following an introduction by the editors, reviews of research will be clustered into two sections. The first part of this Special Issue focuses on the Photonic and RF Communications systems technologies and this Section comprises 7 articles. Recent solutions in wideband and multi-band RF circuits are presented in the articles 7 through 9 in the second part of the Special Issue. The first article, by Arash Bahrami , et al., present an experimental evaluation of a radio-over-fiber system. In the paper, the authors proposed an optical microwave radio-over-fiber system in which an integrated dual-parallel Mach-Zehnder modulator is biased at the maximum transmission biasing point. The authors consider the two following modulation schemes: binary phase shift keying (BSK) and quadrature phase shift keying (QPSK) over fiber spans of 10 and 25 km of standard single mode fiber. The obtained results show that the second order sideband of optical microwave has the potential to provide error-free transmission for BPSK and QPSK. John S. Vardakas , et al. in the next article, analyse the delay of converged Wavelength Division Multiplexing Ethernet Passive Optical Network and WiMax networks. In such access networks, the provision of Quality of Service (QoS) support is a challenging issue, mainly due to the different bandwidth allocation mechanisms of the two access technologies. In the paper, the authors analyse the delay performance by considering multiple service-classes with different priorities. The proposed analytical model allows the calculation of the average end-to-end packet delay of each service-class as the sum of the queuing delay in both domains. The accuracy of the analysis was verified by simulation. Ioannis Mamounakis , et al., propose a novel traffic prediction method to minimize packet delay in Ethernet Passive Optical Networks. The authors utilise traffic information to predict the accumulated burst size of each respective optical network unit in the following cycle. The paper demonstrates and proves that a significant delay enhancement can be accomplished by reporting the predicted burst size, rather than the current one, to the Optical Line Terminal (OLT). The simulation experiments carried out by the authors show that a significant decrease in the delay can be obtained without any changes to the dynamic bandwidth assignment scheme at the OLT. The paper by Mariusz Głąbowski , et al. proposes an analytical model of a multi-service switching network with overflow links with finite capacity in the first stage of the network. This model can be used to evaluate the QoS parameters which are closely interrelated with the method of operation of the large capacity optical nodes. The proposed method allows their traffic capacity to be increased through a significant decrease in the phenomenon of internal blocking in switching networks which is the basic element of the node. The proposed calculation method is validated by the authors via simulation. Aboagela Dogman , et al. present an integrated QoS management process for multimedia traffic in wire and wireless networks. In the proposed process, the statistics of traffic are determined and then used as an input to a fuzzy interface system (FIZ). FIZ allows the traffic to be accurately sampled and this traffic is in the next step pre-processed using a fuzzy c-means (FCM). FIZ identifies clusters representing poor, average and good QoS. This information is then used in a multilayer perceptron neural network to assess the overall QoS for multimedia traffic. The authors showed that the developed adaptive statistical sampling represents traffic more correctly than systematic, stratified and random non-adaptive sampling methods. The article by Adam Kaliszan , et al. provides a generalised convolution algorithm for modelling state-dependent systems. This model corresponds to the system that uses one or many physical resources, e.g. one or a set of lambda channels in DWDM optical networks. The proposed calculation method can be used for the analysis and dimensioning of optical networks and systems. This method is also useful for an optical system in which advanced traffic management mechanism have been applied, e.g. bandwidth reservation or threshold mechanism. The accuracy of the proposed model is validated by simulation experiments. From their onset RF Communications Systems have been critical to the development of ICT infrastructure. In mobile and satellite communication systems, wideband and multi-band RF circuits are important for achieving high-date wireless transmission and supporting a large number of standards (e.g. UMTS, WiFi, WiMAX, or LTE). A wideband or multi-band RF circuit can replace a number of single-band RF circuits, leading to a significant reduction of size, weight and cost of modern wireless systems. There have been a great number of recent developments in the above areas over the past years. The articles that outline these solutions are in the latter part of this Special Issue, i.e. comprise articles from Article 7 through Article 9. The article by Bingo Wing-Kuen Ling , et al. discusses the frequency spectrum of a pulse for ultra wideband impulse radio systems as the frequency response of a real valued causal rational infinite impulse response filter. The pulse design problem is formulated as a functional inequality constrained optimization problem that can be solved by the method proposed by the authors. This pulse can be generated via an excitation of a simple circuit. Numerical simulation results confirm that the energy compaction performance defined in the frequency domain of the proposed pulse outperforms that of existing pulses. It is worth emphasizing that the pulse can be generated via an excitation of a simple circuit. In their article, Xue Li and , et al. present the WiFi leakage problem that appears in the users' equipment that simultaneously use both LTE and WiFi technologies. The authors propose WiFi leakage detection algorithms for successful detection of the weak WiFi leakage in the presence of both LTE signal and Gaussian noise. The algorithms include energy ratio detection, entropy ratio detection and cyclostationary detection based on spectral correlation function. The effectiveness and efficiency of the proposed WiFi leakage detection techniques have been confirmed by simulations performed for different scenarios. The article by Piotr Remlein discusses the performance of Intelligent Transportation Systems (ITS) that consists of satellite mesh networks. In the paper the author analyses a new power-limited and energy-efficient systems employing coded continuous phase modulation signals and with multiuser Frequency Division Multiplexed transmission. The results obtained by the author show that good performance of the proposed solution can be achieved by reducing the inter-carrier frequency spacing and using simple iterative ICI cancellation algorithm at the receiver. Based on the results, it can be concluded that this system might constitute a desirable option in ITS satellite mesh, ad-hoc networks. The objective of this Special Issue is to bring together the state-of-the-art research contributions that address challenges in emerging devices and architecture for photonic systems and RF circuits for wideband and multi-band communications systems. Due to the complexity and scope of the subject area, the set of articles presented attempts to address only some selected relevant issues. The Guest Editors believe though that the contributions included in the Special Issue will inspire the research activities of leading authors and scholars worldwide. The Guest Editors also expect this Special Issue to provide insight into the questions that they address as the presented articles expose important contributions to each of the relevant subject areas. Piotr Zwierzykowski received MSc and PhD (Hons) degrees in telecommunications from Poznan University of Technology, Poland, in 1995 and 2002, respectively. Since 1995 he has been working at the Poznan University of Technology, where he is currently an Assistant Professor. Piotr Zwierzykowski is the author, or co-author, of over 200 journals and conference publications and three books in the area of computer and communication networks. His research covers modelling and dimensioning of optical networks, teletraffic engineering and routing. He is a Senior Member of IEEE and the Secretary in the IEICE Europe Section. Steven Gao received a PhD degree from Shanghai University, P.R. China. He is a Professor and Chair in RF and Microwave Engineering at University of Kent, UK. His research covers smart antennas, satellite antennas, phased arrays, RF and microwave circuits (power amplifiers), satellite communications, UWB radars, synthetic-aperture radars and mobile communications. From 2007 to 2012, he was a Senior Lecturer and Head of Space Antennas and Microwave System Group at Surrey Space Centre, University of Surrey, UK. He has two books including “Space Antenna Handbook” (Wiley, 2012) and “Circularly Polarized Antennas” (IEEE-Wiley, 2014), published over 200 papers and holds several patents in smart antennas and RF. He is an IEEE AP-S Distinguished Lecturer and a Chair of LAPC 2013. He is an Invited Speaker at IWAT'2014 (Sydney), SOMIRES'2013 (Japan), APCAP'2014 (China), etc. He is an Associate Editor of Radio Science and the Editor-in-Chief for Wiley Book Series on "Microwave and Wireless Technologies". He is a Fellow of the IET. Wai Pang Ng received his BEng (Hons) Communication and Electronic Engineering and IEE Prize from Northumbria University, UK and PhD in Electronic Engineering from University of Wales Swansea, UK in 1997 and 2001, respectively. He is a Reader of Optical Communications at Northumbria University. From 2001 to 2004, he was a Senior Networking Software Engineer at Intel Corporation, before starting his academic career at Northumbria University. He has published over 80 journals and conference publications in the area of optical switching, optical signal processing and radio-over-fiber. He is a Chartered Engineer of the Engineering Council UK and a Senior Member of the IEEE. Currently he serves as the Chair of the IEEE UK&RI Communication Chapter. He was also the Co-Chair of the Signal Processing for Communications Symposium in IEEE International Conference on Communications (ICC) 2009 (Dresden, Germany) and now serves as the Publicity Chair of ICC 2015 in London. Zuqing Zhu received a PhD degree from the Department of Electrical and Computer Engineering, University of California, Davis, in 2007. From July 2007 to January 2011, he worked in the Service Provider Technology Group of Cisco Systems, San Jose, as a senior R&D engineer. In January 2011, he joined the University of Science and Technology of China (USTC), where he is currently an Associate Professor. He has published more than 100 papers in peer-reviewed journals and conferences. He has been in the technical program committees (TPC) of INFOCOM, ICC, GLOBECOM, ICCCN and etc. He is also an editorial board member of Elsevier Journal of Optical Switching and Networking, Springer Telecommunication Systems Journal, Springer Photonic Network Communications Journal and Wiley European Transactions on Emerging Telecommunications Technologies amongst others. He has received the Best Paper Awards from the IEEE International Conference on Communications (ICC) 2013, the IEEE Global Communications Conference (GLOBECOM) 2013, and the IEEE International Conference on Networking and Communications (ICNC) 2014. He is a Senior Member of IEEE and OSA.

Travel time uncertainty can have a significant impact on the performance of Intelligent Transportation Systems. Various applications such as routing or pricing can benefit from accurate estimations of travel times. In this thesis, we propose an approach that can estimate travel time distributions for any path in the road network using GPS data. We focus on situations when only sparse data is available: we consider a dataset that contains GPS trajectories from probe vehicles sampled on average with 1-minute intervals. We propose employing a novel Bayesian network inference algorithm and combining it with the Gaussian copula model. This allows us to estimate the marginal distributions and the covariance structure separately.

In this paper, we explore the communication performance of vehicle-to-vehicle (V2V) networks assisted by a reconfigurable intelligent surface (RIS) and a simultaneous transmitting and reflecting intelligent omni-surface (STAR-IOS) under non-orthogonal multiple access (NOMA) and orthogonal multiple access (OMA) schemes. In particular, we consider that the RIS is close to the transmitter vehicle while the STAR-IOS is near the receiver vehicles. In addition, it is assumed that the STAR-IOS exploits the energy-splitting (ES) protocol for communication and the fading channels between the RIS and STAR-IOS follow composite Fisher-Snedecor <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\mathcal {F}$</tex-math></inline-formula> distribution. Under such assumptions, we first use the central limit theorem (CLT) to derive the PDF and the CDF of equivalent channels at receiver vehicles, and then, we derive the closed-form expression of outage probability (OP) under NOMA/OMA scenarios. Additionally, by exploiting Jensen's inequality, we propose an upper bound of the ergodic capacity (EC) and subsequently develop an analytical expression of the energy efficiency (EE) for both NOMA and OMA cases. Further, our analytical results, which are double-checked with the Monte-Carlo simulation, reveal that applying RIS/STAR-RIS in V2V communications significantly enhances the communication performance of intelligent transportation systems (ITS). Besides, the results indicate that considering the NOMA scheme provides better performance in terms of the OP, EC, and EE as compared with the OMA case for the considered V2V communication system.

Globally, buildings are responsible for an estimated 40% of energy consumption and 33% of CO2 emissions. In a bid to reduce CO2 emissions and hence, global warming, it has become necessary to ensure the energy efficient construction and operation of buildings. Understanding how a building utilises energy is a critical step to increase its efficiency. In this study, we leverage on an open-source data obtained from UCI data repository. Exploratory data analysis and feature engineering were used to eliminate non-contributing features while identifying key attributes of the data for model training. Linear Regression (LR) and Support Vector Regression (SVR) were employed as the machine learning techniques for the study. The models were trained using a repeated cross-validation technique. The models’ performance was evaluated on an independent data set segregated for testing. The LR model was trained with nine out of thirty-three features, while the Support Vector Regression (SVR) model used twenty-eight features for its training. The SVR model had a higher variance (0.48), accuracy (92.41%), and lower Mean Absolute Percentage Error (MAPE) of 7.59% compared to the LR model's variance of 0.26, accuracy of 91.87%, and MAPE of 8.13%. The SVR model was more accurate in predicting energy consumption, as it showed better accuracy on the test set with lower MAPE and higher R-squared value. Both models outperformed a relatively complex and computationally expensive model in a previous study. It also identified areas with high energy consumption which could be used to inform the building's energy management strategy.

The current work aims to develop computational models for the thermal characteristics of turbulent CH4 flames for varying burner dimensions. This study develops a platform for data-driven analysis of temperature prediction of turbulent non-premixed flames, in which the influence of flow and geometric parameters, including burner head diameter (D), half cone angles (α), and co-flow air velocity (Ucf), have been considered. The algorithms used were ridge regressor (RR), linear regressor (LR), and three variations of support vector regression (SVR): SVR with a linear kernel (SVR-LR), SVR with a radial basis function (SVR-RBF), and SVR with a polynomial kernel (SVR-Poly). The performance of each computational model was evaluated and contrasted based on several metrics: mean absolute error, regression coefficient (R2), mean absolute percentage error, and mean Poisson deviance. From the modeling of the output data, it was observed that the SVR-RBF predictions were more accurate compared to those from the other algorithms, as it achieved the highest training value of 0.955. The testing predictions of RR, SVR-LR, SVR-RBF, and SVR-Poly algorithms were also robust, with R2 values ranging between 0.91 and 0.94. It is, therefore, established that these computational models are effectively suited for predicting sensitive turbulent CH4 flame characteristics based on varying input factors.

&amp;lt;p&amp;gt;Accurate water level (WL) forecasting is important for water resources management and planning purposes in the Great Lakes. The objectives of this research are two-fold.&amp;amp;#160; The first objective is to apply machine learning (ML) (i.e., random forest (RF) and support vector regression (SVR)) and hybrid convolutional neural network(CNN)-long-short term memory (LSTM) deep learning (DL) models for multi-step (i.e., one-, two- and three-monthly step ahead) WL forecasting in the Great Lakes (Michigan and Ontario). The second objective is to integrate the boundary corrected (BC) maximal overlap discrete wavelet transform (MODWT) with SVR, RF, and CNN-LSTM models to improve the performance of the individual models. By employing a BC-wavelet decomposition method, the &amp;amp;#8216;future data&amp;amp;#8217; issue (i.e., data from the future that is not available), often overlooked in the literature and a major barrier to achieving realistic forecasting performance is overcome.&amp;amp;#160;&amp;lt;/p&amp;gt;&amp;lt;p&amp;gt;For Lakes Michigan and Ontario, 1212 monthly WL (m) records (spanning Jan 1918&amp;amp;#8211;Dec 2018) were used to develop the models. For the non-wavelet-based models (SVR, RF, and CNN-LSTM), candidate model inputs included the WL recorded over the previous 12 months.&amp;amp;#160; For the BC-MODWT-based models (BC-MODWT-SVR, BC-MODWT-RF, and BC-MODWT-CNN-LSTM), the lagged input time series were decomposed into BC-wavelet and scaling coefficients by using different mother wavelets (Haar, Daubechies, Symlets, Fejer-Korovkin and Coiflets), filter lengths (from two up to 12) and decomposition levels (from one up to seven).&amp;amp;#160; For each method (SVR, RF, and CNN-LSTM), mother wavelet, and decomposition level a model was generated.&amp;amp;#160; For both wavelet- and non-wavelet-based models, the particle swarm optimization (PSO) method was used to select the most appropriate inputs to include in the proposed multi-step WL forecasting models.&amp;lt;/p&amp;gt;&amp;lt;p&amp;gt;The datasets were partitioned into calibration and validation subsets. After calibrating the models, various performance evaluation metrics, e.g., coefficient of determination (R&amp;lt;sup&amp;gt;2&amp;lt;/sup&amp;gt;), root mean square error (RMSE), mean absolute error (MAE), root mean square percentage error (RMSPE), mean absolute percentage error (MAPE) and the Nash-Sutcliffe efficiency coefficient (NSC) were used to assess model accuracy.&amp;lt;/p&amp;gt;&amp;lt;p&amp;gt;Of the ML models, the SVR outperformed RF while the DL models outperformed the ML models for each forecast lead time (one-, two-, and three-step(s) ahead). Results from this case study indicate that not all wavelet families and decomposition levels perform equally and in some cases, the wavelet-based models do not improve performance over the non-wavelet-based models. However, the BC-MODWT-CNN-LSTM using suitable mother wavelets (e.g., Haar) outperforms the individual ML and BC-MODWT-ML-based models. More accurate forecasts were obtained for Lake Michigan although the performance in both Great Lakes was accurate. The outcomes of this research indicate that the BC-MODWT-CNN-LSTM model is a promising tool for generating accurate WL forecasts.&amp;lt;/p&amp;gt;