Predicting Graduate Admissions using Ensemble Machine Learning Techniques: A Comparative Study of Classifiers and Regressors

The rapid spread of contagious diseases poses a colossal threat to human existence. Presently, the emergence of coronavirus COVID-19 which has rightly been declared a global pandemic resulting in so many deaths, confusion as well as huge economic losses is a challenge. It has been suggested by the World Health Organization (WHO) in conjunction with different Government authorities of the world and non-governmental organizations, that efforts to curtail the COVID-19 pandemic should rely principally on measures such as social distancing, identification of infected persons, tracing of possible contacts as well as effective isolation of such person(s) for subsequent medical treatment. The aim of this study is to provide a framework for monitoring Movements of Pandemic Disease Patients and predicting their next geographical locations given the recent trend of infected COVID-19 patients absconding from isolation centres as evidenced in the Nigerian case. The methodology for this study, proposes a system architecture incorporating GPS (Global Positioning System) and Assisted-GPS technologies for monitoring the geographical movements of COVID-19 patients and recording of their movement Trajectory Datasets on the assumption that they are assigned with GPS-enabled devices such as smartphones. Accordingly, fifteen (15) participants (patients) were selected for this study based on the criteria of residency and business activity location. The ensuing participants movements generated 157, 218 Trajectory datasets during a period of 3 weeks. With this dataset, mining of the movement trace, Stay Points (hot spots), relationships, and the prediction of the next probable geographical location of a COVID-19 patient was realized by the application of Artificial Intelligence (AI) and Data Mining techniques such as supervised Machine Learning (ML) algorithms (i.e., Multiple Linear Regression (MLR), k-Nearest Neighbor (kNN), Decision Tree Regression (DTR), Random Forest Regression (RFR), Gradient Boosting Regression (GBR), and eXtreme Gradient Boosting regression(XGBR) as well as density-based clustering methods (i.e., DBSCAN) for the computation of Stay Points (hot spots) of COVID-19 patient. The result of this study showed clearly that it is possible to determine the Stay Points (hot spots) of a COVID-19 patient. In addition, this study demonstrated the possibility of predicting the next probable geographical location of a COVID-19 patient. Correspondingly, Six Machine Learning models (i.e., MLR, kNN, DTR, RFR, GBR, and XGBR) were compared for efficiency, in determining the next probable location of a COVID-19 patient. The result showed that the DTR model performed better compared to other models (i.e., MLR, kNN, RFR, GBR, XGBR) based on four evaluation matrices (i.e., ACCURACY, MAE, MSE, and R2) used. It is recommended that less developed Countries consider adopting this framework as a policy initiative for implementation at this burgeoning phase of COVID-19 infection and beyond. The same applies to the developed Countries. There is indication that GPS Trajectory dataset and Machine Learning algorithms as applied in this paper, appear to possess the potential of performing optimally in a real-life situation of monitoring a COVID-19 patient. This paper is unique given its ability to predict the next probable location of a COVID-19 patient. In the review of extant literature, prediction of the next probable location of a COVID-19 patient was not in evidence using the same Machine Learning algorithms.

The shear behaviour of Ultra High-Performance Fiber Reinforced Concrete (UHPFRC) is complex due to fiber contributions alongside conventional reinforcement and concrete effects. This paper presents a novel Machine Learning (ML) approach to predict the shear capacity of UHPFRC beams, offering an alternative to conventional empirical and analytical methods. A database of 282 experimental data points was used to train multiple ML models, including Support Vector Regression (SVR), k-Nearest Neighbour (kNN), Decision Tree Regressor (DTR), Random Forest Regressor (RFR), Artificial Neural Network (ANN), Gradient Boosting Regressor (GBR), Light Gradient Boosting Regressor (LGBR), and Extreme Gradient Boosting Regressor (XGBR). Among these, boosting-based models performed best, with XGBR achieving the highest accuracy with an R 2 score of 0.91. Further, a feature importance analysis identified the shear span-to-depth ratio as the most influential parameter on the shear capacity, with longitudinal and transverse reinforcement ratios also playing key roles. A SHapley Additive exPlanation (SHAP) analysis was conducted, offering an interpretable framework for understanding the contribution of each feature to the model’s prediction. This analysis further clarified feature contributions, identifying optimal parameter ranges for shear capacity. This data-driven approach bridges empirical methods and computational modelling enhancing confidence in ML-based predictions and guiding efficient UHPFRC design.

ObjectiveThis study aims to develop and compare different models to predict the Length of Stay (LoS) and the Prolonged Length of Stay (PLoS) of inpatients admitted through the emergency department (ED) in general patient settings. This aim is not only to promote any specific model but rather to suggest a decision-supporting tool (i.e., a prediction framework).MethodsWe analyzed a dataset of patients admitted through the ED to the “Sant”Orsola Malpighi University Hospital of Bologna, Italy, between January 1 and October 26, 2022. PLoS was defined as any hospitalization with LoS longer than 6 days. We deployed six classification algorithms for predicting PLoS: Random Forest (RF), Support Vector Machines (SVM), Gradient Boosting (GB), AdaBoost, K-Nearest Neighbors (KNN), and logistic regression (LoR). We evaluated the performance of these models with the Brier score, the area under the ROC curve (AUC), accuracy, sensitivity (recall), specificity, precision, and F1-score. We further developed eight regression models for LoS prediction: Linear Regression (LR), including the penalized linear models Least Absolute Shrinkage and Selection Operator (LASSO), Ridge and Elastic-net regression, Support vector regression, RF regression, KNN, and eXtreme Gradient Boosting (XGBoost) regression. The model performances were measured by their mean square error, mean absolute error, and mean relative error. The dataset was randomly split into a training set (70%) and a validation set (30%).ResultsA total of 12,858 eligible patients were included in our study, of whom 60.88% had a PloS. The GB classifier best predicted PloS (accuracy 75%, AUC 75.4%, Brier score 0.181), followed by LoR classifier (accuracy 75%, AUC 75.2%, Brier score 0.182). These models also showed to be adequately calibrated. Ridge and XGBoost regressions best predicted LoS, with the smallest total prediction error. The overall prediction error is between 6 and 7 days, meaning there is a 6–7 day mean difference between actual and predicted LoS.ConclusionOur results demonstrate the potential of machine learning-based methods to predict LoS and provide valuable insights into the risks behind prolonged hospitalizations. In addition to physicians' clinical expertise, the results of these models can be utilized as input to make informed decisions, such as predicting hospitalizations and enhancing the overall performance of a public healthcare system.

The growth of every nation's economy depends heavily on the stock market. Because of their strong predictive powers, "K-Nearest Neighbor Regression (KNNR), Random Forest Regression (RFR), Decision Tree Regression (DTR), and Support Vector Regression (SVR)" is commonly used. From the previous study, it shows that the chosen state-of-the-art of previous studies is by Shah et al. (2021) using Bi-LSTM. But the state-of-the-art model shows a high Mean Squared Error (MSE) value that is 5% for predict Tesla Stock. Therefore, we propose the SVR, KNNR, DTR, and RFR model hope error value can be reduced. However, those models have difficulties finding suitable parameters for prediction. Due to that, we proposed the best parameters that were commonly used in studies which is SVR with RBF kernel, C = [1; 10; 100], gamma = [0, 1; 0, 2], epsilon = 0,1, KNNR with K = [1, 3, 5, 7, 9], and DTR with a criterion = [‘squared_error’;’friedman_mse’], RFR with 0 < n_estimators < 100 is picked to enhance the predictive capabilities of each proposed model. To prove this, a comparison is made with each proposed model, and choose the best proposed model then compared to the state-of-the-art of the previous study, which is Bi-LSTM. The dataset used was the Tesla Stock historical data from yahoo finance. The result showed RFR with n_estimators = 87 is superior to its comparison with results of 0.005452 RMSE, 0.000029 MSE, and 0.999296 R2 and compared to Bi-LSTM have about 0.11773 RMSE, 0.01386 MSE, and 0.791263 R2. Based on this study, it can be concluded that RFR with n_estimators = 87 has superior predictive capabilities performance compared with other models and the state-of-the-art of previous studies. Doi: 10.28991/HEF-2022-03-04-01 Full Text: PDF

This paper presents a novel application of machine learning models to clarify the intricate behaviors of expansive soils, focusing on the impact of sand content, saturation level, and dry density. Departing from conventional methods, this research utilizes a data-centric approach, employing a suite of sophisticated machine learning models to predict soil properties with remarkable precision. The inclusion of a 30% sand mixture is identified as a critical threshold for optimizing soil strength and stiffness, a finding that underscores the transformative potential of sand amendment in soil engineering. In a significant advancement, the study benchmarks the predictive power of several models including extreme gradient boosting (XGBoost), gradient boosting regression (GBR), random forest regression (RFR), decision tree regression (DTR), support vector regression (SVR), symbolic regression (SR), and artificial neural networks (ANNs and proposed ANN-GMDH). Symbolic regression equations have been developed to predict the elasticity modulus and unconfined compressive strength of the investigated expansive soil. Despite the complex behaviors of expansive soil, the trained models allow for optimally predicting the values of unconfined compressive parameters. As a result, this paper provides for the first time a reliable and simply applicable approach for estimating the unconfined compressive parameters of expansive soils. The proposed ANN-GMDH model emerges as the pre-eminent model, demonstrating exceptional accuracy with the best metrics. These results not only highlight the ANN’s superior performance but also mark this study as a groundbreaking endeavor in the application of machine learning to soil behavior prediction, setting a new benchmark in the field.

As Monkeypox becomes a prevalent public health issue, it is important to develop advanced detection and prediction methods that will inform public health strategies that govern Monkeypox prevention. This study employs machine learning methods to analyze and predict Monkeypox case trends. In particular, features on new cases and deaths were applied to regression and classification models to predict the total number of Monkeypox cases and new case probablity. The regression models that were applied included Linear regression (LR), Decision Tree Regression (DT), Random Forest Regression (RF), Support Vector Regression (SVR), and K-Nearest Neighbor Regression (KNN), with total cases as the outcome. Among regression methods, the Random Forest Regression model performed the best with a Mean Squared Error (MSE) of 92,425,437.81 and R-squared of 0.06, indeicating moderate predictive ability. The methods were also similar to predict new cases, and once again the same algorithms were applied to classification methods, including Decision Tree (DT), Random Forest (RF), and K-Nearest Neighbor (KNN) classification, and each model achieved an accuracy score of one (1.00), indicating no new cases would be missed. These results provide evidence that these are effective machine learning methods, and random forests in particular provides the best predictive capability for Monkeypox case trend analysis. The results illustrate how these models can assist data-driven decisions in public health, and evidence-based preparedness and response for future Monkeypox outbreaks.

Personal identification of an individual has always been a major concern in forensic science. Reconstruction of the facial profile is considered as one of the final stages in the process of identification. Nevertheless, recent advancements in artificial intelligence (AI) and machine learning (ML) have demonstrated remarkable potential in predictive modelling and forensic applications. The current study uses customised machine learning models to predict facial dimensions based on dental and jaw parameters. A sample of 422 participants (201 males and 221 females) from a North Indian population was collected and analysed. Dental casts, anthropometric facial measurements and photographs of the participants were collected with informed consent. ML models such as Support Vector Regression (SVR), Random Forest Regression (RFR), Decision Tree Regression (DTR), and Linear Regression (LR) were trained using dental and jaw measurements as input features for the models. The results show that the ML models predicted the facial dimensions with an accuracy of 90–94% and a very low prediction error of 0.1–0.9 across all facial measurements. Among the models, SVR and LR models perform well, followed by RFR, whereas DFR yielded comparatively lower accuracy. The findings demonstrate that machine learning models (SVR, RFR, DTR, and LR) can be used as novel approach to predict facial dimensions from jaw and teeth parameters. These techniques can be combined with other facial reconstruction techniques to produce more precise and accurate outcomes. The reliability and accuracy in predicting the facial dimensions indicate that the results can be applied in the practical and real situations such as personal identification, forensic investigations, disaster victim identification cases, and archaeological remains where only jaw and teeth are available for examination. Integrating ML-based predictions with traditional facial reconstruction techniques could enhance the accuracy and reliability of forensic identification methodologies.

This work evaluates machine learning (ML) and deep learning (DL) models for Photovoltaic (PV) power output prediction based on two real-world datasets. Eight models – Support Vector Regression (SVR), K-Nearest Neighbors (KNN), Linear Regression (LR), Random Forest Regression (RFR), Gradient Boosting Regression (GBR), Decision Tree (DT), one Dimensional Convolutional Neural Network (1DCNN) and Artificial Neural Network (ANN) – were evaluated using Root Mean Squared Error (RMSE), R-squared (R²), Mean Absolute Error (MAE), and Mean Squared Error (MSE). For Dataset1 (region of Cacak) 1DCNN achieved the optimal performance, with the reduced errors and the largest R² values at three spatial locations, and it was closely followed by RFR and GBR. Likewise, in Dataset2 (region of Kraljevo), 1DCNN, ANN, and RFR give the best results. Conventional models, such as LR and DT, performed poorly in both data tribes. The results highlight the ability of state-of-the-art ensemble and DL models in learning nonlinear patterns in solar energy data and thus, the importance of choosing the right prediction tools to provide accurate and reliable PV power forecasts.  

For forecasting the spread of dengue, monitoring climate change and its effects specific to the disease is necessary. Dengue is one of the most rapidly spreading vector-borne infectious diseases. This paper proposes a forecasting model for predicting dengue incidences considering climatic variability across nine cities of Maharashtra state of India over 10 years. The work involves the collection of five climatic factors such as mean minimum temperature, mean maximum temperature, relative humidity, rainfall, and mean wind speed for 10 years. Monthly incidences of dengue for the same locations are also collected. Different regression models such as random forest regression, decision trees regression, support vector regress, multiple linear regression, elastic net regression, and polynomial regression are used. Time-series forecasting models such as holt's forecasting, autoregressive, Moving average, ARIMA, SARIMA, and Facebook prophet are implemented and compared to forecast the dengue outbreak accurately. The research shows that humidity and mean maximum temperature are the major climate factors and exhibit strong positive and negative correlation, respectively, with dengue incidences for all locations of Maharashtra state. Mean minimum temperature and rainfall are moderately positively correlated with dengue incidences. Mean wind speed is a less significant factor and is weakly negatively correlated with dengue incidences. Root mean square error (RMSE), mean absolute error (MAE), and R square error (R2) evaluation metrics are used to compare the performance of the prediction model. Random Forest Regression is the best-fit regression model for five out of nine cities, while Support Vector Regression is for two cities. Facebook Prophet Model is the best fit time series forecasting model for six out of nine cities. Based on the prediction, Mumbai, Thane, Nashik, and Pune are the high-risk regions, especially in August, September, and October. The findings exhibit an effective early warning system that would predict the outbreak of other infectious diseases. It will help the relevant authorities to take accurate preventive measures.

Purpose The purpose of this study is to select the appropriate forecasting model at the retail stage for selected vegetables on the basis of performance analysis. Design/methodology/approach Various forecasting models such as the Box–Jenkins-based auto-regressive integrated moving average model and machine learning-based algorithms such as long short-term memory (LSTM) networks, support vector regression (SVR), random forest regression, gradient boosting regression (GBR) and extreme GBR (XGBoost/XGBR) were proposed and applied (i.e. modeling, training, testing and predicting) at the retail stage for selected vegetables to forecast demand. The performance analysis (i.e. forecasting error analysis) was carried out to select the appropriate forecasting model at the retail stage for selected vegetables. Findings From the obtained results for a case environment, it was observed that the machine learning algorithms, namely LSTM and SVR, produced the better results in comparison with other different demand forecasting models. Research limitations/implications The results obtained from the case environment cannot be generalized. However, it may be used for forecasting of different agriculture produces at the retail stage, capturing their demand environment. Practical implications The implementation of LSTM and SVR for the case situation at the retail stage will reduce the forecast error, daily retail inventory and fresh produce wastage and will increase the daily revenue. Originality/value The demand forecasting model selection for agriculture produce at the retail stage on the basis of performance analysis is a unique study where both traditional and non-traditional models were analyzed and compared.

Indoor temperature is an important criterion for evaluating the operation quality of district heating systems (DHSs) and has a significant impact on improving the energy efficiency of heat exchange station regulation. Accurate prediction of indoor temperature is of great help to the precise control of heating systems. However, due to the thermal inertia and response delay characteristics of heating systems, there is a complex nonlinear relationship between indoor temperature and its many related factors. A hierarchical attention gated recurrent unit (HAGRU) neural network is innovatively proposed for predicting the indoor temperature of energy-saving buildings in which the indoor temperature is optimally regulated based on a newly designed smart on-off valve. The network is divided into two levels of attention model, which can realize the representation of a single influencing factor and the fusion of multiple feature inputs. Detailed simulation results show that the predictive accuracy of the proposed algorithm is 98.4%, which is significantly better than state-of-art algorithms, such as support vector machine (SVM), random forest regression (RFR), decision tree regression (DTR), gradient boosting regression (GBR), recurrent neural network (RNN), long short-term memory (LSTM) and gated recurrent units (GRU). Therefore, the proposed HAGRU algorithm has good nonlinear feature extraction and expression ability. The indoor temperature prediction results are used as feedback input to assist the control strategy of the heat exchanger station, which is conducive to the improvement of energy efficiency. In addition, the optimal selection of hyperparameters of the HAGRU algorithm is also analyzed in detail.

The early detection and rapid identification of life-threatening diseases play crucial roles in saving lives. However, disparities in healthcare infrastructure and the unequal distribution of medical resources often hinder timely diagnosis, resulting in adverse outcomes. This study introduces a Multi-Disease Prediction System, a user-friendly web application developed with Streamlit, capable of predicting kidney disease, breast cancer, and heart disease. Unlike traditional AI models that focus on predicting a single disease, our objective is to develop a unified platform using Streamlit and Machine learning to predict multiple diseases simultaneously. The system harnesses advanced machine learning techniques, utilizing algorithms such as the Random Forest, Gradient Boosting Classifier, XGBoost, K-Nearest Neighbor, and Decision Tree classifier. For each disease, the optimal algorithm was selected based on its performance, yielding predictive accuracies of 99% for kidney disease (Random Forest Classifier), 98% for breast cancer (XGBoost), and 95% for heart disease (Random Forest Classifier). Users can input health indicators and symptoms and receive personalized risk insights. By offering an intuitive and interactive platform for data-driven screening, this application seeks to enhance early detection and encourage proactive health management in diverse populations. These models are designed to continuously improve through advancements in machine learning, ensuring reliable and impactful health predictions for broader accessibility and preventive care. Additionally, future improvements, such as integrating more diseases into the system, would further enhance its capabilities and expand its impact on early disease detection.

International Dark-Sky Places (IDSPs) is a global project led by the International Dark-Sky Association (IDA) to protect dark-sky areas for astronomical observation and astro-tourism. Currently, there are 216 IDSPs in the world. This study analyses light pollution trends in International Dark-Sky Places (IDSPs) using VIIRS night light composite data. IDSPs are mostly located near urbanised areas in Europe and away from urbanisation in North America. Dark and large areas away from urbanisation in South America, Africa, Central Asia and Australia offer great potential. Linear regression analysis shows that in 94% out of 216 places there is no change in artificial light at nigh (ALAN). There is a decrease in the level of artificial light in 3% and an increase in the level of artificial light in 3%. Unlike the linear regression analysis, Mann–Kendall and Sen slope analyses revealed significant statistical changes at the 105 sites (78 increasing and 27 decreasing). Machine learning models including decision tree regression (DTR), random forest regression (RFR) and gradient boosting regression (GBR) provided good predictive accuracy (R $$^{2}$$ 2 > 0.6). Seasonal patterns were assessed using the Lomb–Scargle periodogram, but detected periodicities were considered false positives. These results emphasise the need for strategic planning to reduce light pollution in IDSPs, especially in underused areas.

Transmission line failures pose significant risks to the reliability and safety of electrical power systems, often resulting in costly outages and prolonged system downtimes. Early and accurate detection of potential failures is crucial to ensure uninterrupted power delivery and facilitate efficient maintenance planning. In this project, we present a robust classification framework for transmission line failure prediction using ensemble machine learning techniques, with a focus on enhancing predictive accuracy and model robustness. The study begins with comprehensive Exploratory Data Analysis (EDA) to uncover underlying patterns and relationships within the dataset. Using statistical summaries, correlation analysis, and visualization techniques, key features influencing transmission line failures are identified. This analysis is instrumental in refining the dataset and improving model performance. A Decision Tree Classifier is initially employed as the baseline model due to its simplicity and interpretability. However, it demonstrates limitations such as overfitting and poor generalization on complex or noisy data. To address these challenges, we propose a Random Forest Classifier as the improved model. By aggregating predictions from multiple decision trees, the Random Forest significantly enhances classification accuracy, mitigates overfitting, and better captures feature interactions. Experimental results show that the Random Forest Classifier consistently outperforms the Decision Tree across all major performance metrics, confirming its effectiveness in accurately predicting transmission line failures. This project underscores the value of ensemble methods in critical infrastructure applications and demonstrates how data-driven approaches can be leveraged to build intelligent fault detection systems in the power sector.

Air pollution is a leading cause of mortality in the developing world, with industrial regions experiencing particularly high levels of fine particulate matter (PM2.5). The Assam Oil Field Region, a significant industrial hub, is characterized by elevated PM2.5 concentrations, yet predictive modeling studies in such environments remain scarce. This study presents a preliminary exploration of statistical and machine learning techniques, aiming to identify potential tools for future environmental monitoring, in oil field localities. Using a limited dataset (n = 109), we modelled PM2.5 levels based on multiple predictor variables, including meteorological parameters (temperature, humidity, wind speed, and wind direction). We employed multiple linear regression alongside six machine learning models—Linear Regression, Decision Tree Regression, Support Vector Regression, KNN Regression, XGBoost Regression (XGBR), and Random Forest Regression, and one Feed Forward Deep Learning model. Additionally, Principal Component Analysis was employed to investigate the impact of dimensionality reduction on model performance. Among the models, XGBR consistently demonstrated superior predictive performance, achieving high R2 scores (0.78 in non-PCA, 0.59 in PCA) and minimal error metrics (MAE: 6.75, MSE: 62.72, RMSE: 7.92). Its robustness across scenarios and ability to handle complex relationships efficiently make it particularly effective compared to other ensemble methods like Random Forest Regression. While initial outcomes suggest the potential of machine learning models, especially XGBoost, for PM2.5 predictions in industrial contexts, further validation using larger and more diverse datasets is necessary. These early findings may serve as a basis for future work on data-driven air quality forecasting in oil field regions.