Ensemble Prediction Method Research Articles

Implementation of Ensemble Learning Algorithm - Stacking Regressor on PM2.5 Prediction Model

This study aims to develop a PM2.5 concentration prediction model using ensemble learning techniques, focusing on the application of stacking regressor. The developed model is compared with several other basic models, namely LSTM, Random Forest, XGBoost, and GBM, to evaluate its performance in terms of prediction accuracy. The results show that the stacking regressor model provides more accurate prediction results than these basic models. The LSTM model has a Mean Squared Error (MSE) value of 80.42 and a coefficient of determination (R²) of -0.29, which shows unsatisfactory performance despite being able to capture temporal patterns. Random Forest gave better results with an MSE of 10.90 and R² of 0.83, thanks to its ability to handle heterogeneous data. The XGBoost and GBM models showed similar performance, with MSE of 9.01 and 8.81 respectively and R² of 0.86. However, the stacking regressor model with the RidgeCV meta-learner achieved the best results with an MSE of 8.07 and R² of 0.87, indicating that this technique successfully combines the advantages of various base models to improve prediction accuracy. In conclusion, the stacking regressor model proved effective in improving PM2.5 prediction accuracy, making it a potential tool for air quality monitoring and supporting public health policy making. This research makes an important contribution to the development of ensemble learning-based air quality prediction methods, and the results can serve as a reference for future research

Fast and scalable ensemble learning method for versatile polygenic risk prediction

Polygenic risk scores (PRS) enhance population risk stratification and advance personalized medicine, but existing methods face several limitations, encompassing issues related to computational burden, predictive accuracy, and adaptability to a wide range of genetic architectures. To address these issues, we propose Aggregated L0Learn using Summary-level data (ALL-Sum), a fast and scalable ensemble learning method for computing PRS using summary statistics from genome-wide association studies (GWAS). ALL-Sum leverages a L0L2 penalized regression and ensemble learning across tuning parameters to flexibly model traits with diverse genetic architectures. In extensive large-scale simulations across a wide range of polygenicity and GWAS sample sizes, ALL-Sum consistently outperformed popular alternative methods in terms of prediction accuracy, runtime, and memory usage by 10%, 20-fold, and threefold, respectively, and demonstrated robustness to diverse genetic architectures. We validated the performance of ALL-Sum in real data analysis of 11 complex traits using GWAS summary statistics from nine data sources, including the Global Lipids Genetics Consortium, Breast Cancer Association Consortium, and FinnGen Biobank, with validation in the UK Biobank. Our results show that on average, ALL-Sum obtained PRS with 25% higher accuracy on average, with 15 times faster computation and half the memory than the current state-of-the-art methods, and had robust performance across a wide range of traits and diseases. Furthermore, our method demonstrates stable prediction when using linkage disequilibrium computed from different data sources. ALL-Sum is available as a user-friendly R software package with publicly available reference data for streamlined analysis.

A coarse-to-fine ensemble method for capacity prediction of lithium-ion batteries in production

ABSTRACT The grading process is a critical stage in the production of lithium-ion batteries. Measuring capacity by full discharge is time-consuming and energy-intensive. Therefore, accurate and efficient capacity prediction is essential. However, current methods still require improvements in data selection, prediction accuracy and handling of dirty data. This paper proposes a coarse-to-fine ensemble learning framework using LightGBM regression algorithm to predict battery capacity. The framework uses raw statistical data directly, instead of complicated features extraction. From different perspectives of the whole data and classified data, the capacity is predicted from coarse to fine, and the semi-dirty data is predicted separately. The method is verified using lithium-ion battery datasets collected from actual industrial production lines. The experimental results show that the mean absolute error and mean absolute percentage error on the test set are 0.31 Ah and 0.11%, respectively, outperforming other regression prediction algorithms. Additionally, semi-dirty data, which accounts for 0.26% of the dataset, is predicted effectively.

A novel ensemble deep reinforcement learning model for short‐term load forecasting based on Q‐learning dynamic model selection

AbstractShort‐term load forecasting is critical for power system planning and operations, and ensemble forecasting methods for electricity loads have been shown to be effective in obtaining accurate forecasts. However, the weights in ensemble prediction models are usually preset based on the overall performance after training, which prevents the model from adapting in the face of different scenarios, limiting the improvement of prediction performance. In order to improve the accurateness and validity of the ensemble prediction method further, this paper proposes an ensemble deep reinforcement learning approach using Q‐learning dynamic weight assignment to consider local behaviours caused by changes in the external environment. Firstly, the variational mode decomposition is used to reduce the non‐stationarity of the original data by decomposing the load sequence. Then, the recurrent neural network (RNN), long short‐term memory (LSTM), and gated recurrent unit (GRU) are selected as the basic power load predictors. Finally, the optimal weights are ensembled for the three sub‐predictors by the optimal weights generated using the Q‐learning algorithm, and the final results are obtained by combining their respective predictions. The results show that the forecasting capability of the proposed method outperforms all sub‐models and several baseline ensemble forecasting methods.

Verification of Long‐Term Ensemble Evapotranspiration Hindcast Using a Conditional Nonlinear Optimal Parameter Perturbation Ensemble Prediction Method on the Tibetan Plateau

AbstractIn this research, a conditional nonlinear optimal perturbation related to parameters (CNOP‐P) method is employed to propose an ensemble prediction method titled as the conditional nonlinear optimal perturbation related to parameters ensemble prediction (CNOP‐PEP) method. Within the CNOP‐PEP method, all ensemble members are generated by the CNOP‐P method. To explore the operability and validity of the CNOP‐PEP method, long‐term evapotranspiration (ET) hindcast skill is evaluated at 13 sites on the Tibetan Plateau (TP) during the period of 2001–2018 with a land surface model. Two traditional ensemble prediction methods (the stochastically perturbed parameters (SPP) method and the one‐at‐a‐time (OAT) method) are also applied to assess the ET hindcast skills. The numerical results indicate that the CNOP‐PEP method shows the best hindcast skill for estimated ET at 13 stations on the TP among the three methods (CNOP‐PEP, OAT, and SPP methods) during the period of 2001–2018. This suggests that the CNOP‐PEP method is helpful and effective for long‐term and interannual hydrological predictions, such as long‐term and interannual ET predictions. The numerical results also indicate that ensemble members with certain special properties of fast‐growing types should be selected to implement ensemble prediction compared to these ordinary and traditional samples. The CNOP‐PEP method could supply special samples to improve the ensemble prediction and hindcast skill of hydrological cycle.

Signature-based Adaptive Cloud Resource Usage Prediction Using Machine Learning and Anomaly Detection

One of the challenges in managing cloud computing clusters is assigning resources based on the customers’ needs. For this mechanism to work efficiently, it is imperative that there are sufficient resources reserved to maintain continuous operation, but not too much to avoid overhead costs. Additionally, to avoid the overhead of acquisition time, it is important to reserve resources sufficiently in advance. This paper presents a novel reliable general-purpose mechanism for prediction-based resource usage reservation. The proposed solution should be capable of operating for long periods of time without drift-related problems, and dynamically adapt to changes in system usage. To achieve this, a novel signature-based ensemble prediction method is presented, which utilizes multiple distinct prediction algorithms suited for various use-cases, as well as an anomaly detection mechanism used to improve prediction accuracy. This ensures that the mechanism can operate efficiently in different real-life scenarios. Thanks to a novel signature-based selection algorithm, it is possible to use the best available prediction algorithm for each use-case, even over long periods of time, which would typically lead to drifts. The proposed approach has been evaluated using real-life historical data from various production servers, which include traces from more than 1,500 machines collected over more than a year. Experimental results have demonstrated an increase in prediction accuracy of up to 21.4 percent over the neural network approach. The evaluation of the proposed approach highlights the importance of choosing the appropriate prediction method, especially in diverse scenarios where the load changes frequently.

An Ensemble of the Convolutional Neural Network Model with Fuzzy Fusion Rank Algorithm for the Identification of Parkinson’s Disease Using Magnetic Resonance Imaging Images

Abstract Background: Parkinson’s disease (PD) is a degenerative condition of the central nervous system primarily affecting the substantia nigra in the brain, resulting in the loss of dopamine-producing neurons and subsequent motor function deterioration. Early symptoms typically include hand tremors followed by bradykinesia, rigidity, postural instability, and decreased balance. Early diagnosis and proactive management are crucial for improving patients’ quality of life. Methods: In this study, an ensemble of deep learning (DL) models was developed to predict PD using DaTscan images. Initially, DL models (VGG16, ResNet50, Inception-V3 were utilized to classify PD in its early stage. Subsequently, to enhance the overall performance of the classification model, an ensemble strategy based on the fuzzy fusion rank algorithm was employed. The Parkinson’s Progression Markers Initiative database served as the evaluation dataset for the proposed model. This may offer some insight into why certain involuntary symptoms of PD occur, such as fatigue, irregular blood pressure, diminished peristalsis, and unexpected decreases in blood pressure. The ensemble model demonstrated superior recognition accuracy, precision, sensitivity, specificity, and F1-score, achieving 98.45%, 98.84%, 98.84%, 97.67%, and 98.84%, respectively. Results: Compared to individual models, the ensemble model demonstrated superior recognition accuracy, precision, sensitivity, specificity, and F1-score, achieving 98.92%, 98.84%, 98.84%, 97.67%, and 98.84%, respectively. In addition, a publicly available graphical user interface (GUI)-based software program was developed, leveraging magnetic resonance imaging for efficient and accurate classification of PD and its subclasses. The method outperformed contemporary techniques in PD detection, offering significant potential for real-time disease identification. Conclusion: The establishment of a GUI-based software application can substantially aid in the timely detection of PD, facilitating proactive management and improving patient outcomes. The study underscores the importance of leveraging advanced DL techniques and ensemble methods for accurate and efficient disease prediction. Moreover, the development of user-friendly software tools holds promise for widespread adoption and enhanced patient care in the field of PD diagnosis and management.

A stacking-based ensemble prediction method for multiobjective aerodynamic optimization of high-speed train nose shape

A stacking-based ensemble prediction method is proposed in order to improve the efficiency and accuracy of aerodynamic shape optimization. This can be divided into a two-level model. The first-level model uses various base learners to predict the training dataset and obtain various combinations of predictions. In the second-level model, the meta-learner is trained using various combinations of predictions as inputs and the true response values as outputs. The RMSE and R-Square metrics results show that the performance metrics of the stacked models are significantly better than those of the original surrogate models, except for the stacked RBFNN model. The stacked Kriging model and LSSVR model achieved the best results with RMSE and R-Square index values of 0.0039 and 0.8418 for aerodynamic drag coefficient, respectively. The stacked Kriging model, RBFNN model, and LSSVR model achieved the best results with RMSE and R-Square index values of 0.0024 and 0.76 for aerodynamic lift coefficient, respectively. Four state-of-the-art multiobjective optimization algorithms are used to perform the optimization process. The results of hypervolume and Pareto fronts demonstrate the effectiveness of the selected FDV-NSGAII algorithm. After optimization, the drag and lift coefficients before and after optimization are reduced by 1.9% and 12.7%, respectively.

An ensemble learning method for Bitcoin price prediction based on volatility indicators and trend

Predicting the price of Bitcoin poses a challenge for researchers, merchants, traders and investors alike. This paper delves into the analysis of a Bitcoin price and volume dataset, spanning from September 2014 to July 2023. The objective is to extract multiple features related to price volatility and employ them to forecast the Bitcoin price for the subsequent 7 days. To achieve this, an Ensemble Learning Method (ELM) is proposed, able to estimate prices in both bullish and bearish markets. For price prediction, we consider three categories of predictors: 1) Bitcoin historical data; 2) volatility indicators; 3) trend prediction (price up or down) obtained through binary classification. Further, we employ a combination of ensemble models (regressors and classifiers) to predict the price at the daily level. The predictions of these models are stacked and weighted by the proposed ELM to improve the forecast accuracy. The ELM is rigorously tested under various market scenarios, yielding results that demonstrate a noteworthy level of forecast accuracy. The period of 2021 stands out as particularly interesting for prediction due to several dramatic price swings. The ELM achieves a substantial 26% improvement in overall accuracy compared to the best-performing individual ensemble model. Throughout the entire year-2021, the Mean Absolute Error (MAE) stood at 319 USD, indicating a notably low MAE.

Ensemble Prediction Method Based on Decomposition–Reconstitution–Integration for COVID-19 Outbreak Prediction

Due to the non-linear and non-stationary nature of daily new 2019 coronavirus disease (COVID-19) case time series, existing prediction methods struggle to accurately forecast the number of daily new cases. To address this problem, a hybrid prediction framework is proposed in this study, which combines ensemble empirical mode decomposition (EEMD), fuzzy entropy (FE) reconstruction, and a CNN-LSTM-ATT hybrid network model. This new framework, named EEMD-FE-CNN-LSTM-ATT, is applied to predict the number of daily new COVID-19 cases. This study focuses on the daily new case dataset from the United States as the research subject to validate the feasibility of the proposed prediction framework. The results show that EEMD-FE-CNN-LSTM-ATT outperforms other baseline models in all evaluation metrics, demonstrating its efficacy in handling the non-linear and non-stationary epidemic time series. Furthermore, the generalizability of the proposed hybrid framework is validated on datasets from France and Russia. The proposed hybrid framework offers a new approach for predicting the COVID-19 pandemic, providing important technical support for future infectious disease forecasting.

A Novel Heuristic-Based Selective Ensemble Prediction Method for Digital Financial Fraud Risk

A Novel Heuristic-Based Selective Ensemble Prediction Method for Digital Financial Fraud Risk

Towards robust seasonal streamflow forecasts in mountainous catchments: impact of calibration metric selection in hydrological modeling

Abstract. Dynamical (i.e., model-based) methods are widely used by forecasting centers to generate seasonal streamflow forecasts, building upon process-based hydrological models that require parameter specification (i.e., calibration). Here, we investigate the extent to which the choice of calibration objective function affects the quality of seasonal (spring–summer) streamflow hindcasts produced with the traditional ensemble streamflow prediction (ESP) method and explore connections between hindcast skill and hydrological consistency – measured in terms of biases in hydrological signatures – obtained from the model parameter sets. To this end, we calibrate three popular conceptual rainfall-runoff models (GR4J, TUW, and Sacramento) using 12 different objective functions, including seasonal metrics that emphasize errors during the snowmelt period, and produce hindcasts for five initialization times over a 33-year period (April 1987–March 2020) in 22 mountain catchments that span diverse hydroclimatic conditions along the semiarid Andes Cordillera (28–37∘ S). The results show that the choice of calibration metric becomes relevant as the winter (snow accumulation) season begins (i.e., 1 July), enhancing inter-basin differences in hindcast skill as initializations approach the beginning of the snowmelt season (i.e., 1 September). The comparison of seasonal hindcasts shows that the hydrological consistency – quantified here through biases in streamflow signatures – obtained with some calibration metrics (e.g., Split KGE (Kling–Gupta efficiency), which gives equal weight to each water year in the calibration time series) does not ensure satisfactory seasonal ESP forecasts and that the metrics that provide skillful ESP forecasts (e.g., VE-Sep, which quantifies seasonal volume errors) do not necessarily yield hydrologically consistent model simulations. Among the options explored here, an objective function that combines the Kling–Gupta efficiency (KGE) and the Nash–Sutcliffe efficiency (NSE) with flows in log space provides the best compromise between hydrologically consistent simulations and hindcast performance. Finally, the choice of calibration metric generally affects the magnitude, rather than the sign, of correlations between hindcast quality attributes and catchment descriptors, the baseflow index and interannual runoff variability being the best predictors of forecast skill. Overall, this study highlights the need for careful parameter estimation strategies in the forecasting production chain to generate skillful forecasts from hydrologically consistent simulations and draw robust conclusions on streamflow predictability.

Ensemble methods for testing a global null

Abstract Testing a global null is a canonical problem in statistics and has a wide range of applications. In view of the fact that no uniformly most powerful test exists, prior and/or domain knowledge are commonly used to focus on a certain class of alternatives to improve the testing power. However, it is generally challenging to develop tests that are particularly powerful against a certain class of alternatives. In this paper, motivated by the success of ensemble learning methods for prediction or classification, we propose an ensemble framework for testing that mimics the spirit of random forests to deal with the challenges. Our ensemble testing framework aggregates a collection of weak base tests to form a final ensemble test that maintains strong and robust power for global nulls. We apply the framework to four problems about global testing in different classes of alternatives arising from whole-genome sequencing (WGS) association studies. Specific ensemble tests are proposed for each of these problems, and their theoretical optimality is established in terms of Bahadur efficiency. Extensive simulations and an analysis of a real WGS dataset are conducted to demonstrate the type I error control and/or power gain of the proposed ensemble tests.

A weight-allocation-based ensemble remaining useful life prediction approach for a single device

Accurate prediction of the remaining useful life (RUL) can improve device reliability and reduce maintenance costs. In real working conditions, the absence of historical data for devices of the same type and the unlabeled degradation data for a single device present significant challenges for predicting RUL. Furthermore, it is difficult to predict RUL accurately due to the randomness and unstable prediction accuracy of an individual model in different scenarios. To address these issues, this paper proposes an ensemble prediction method based on weight allocation. Consider the advantages and disadvantages of various prediction methods, Particle Filter (PF), Support Vector Regression (SVR), and Gaussian Process Regression (GPR) are selected as member models for the ensemble prediction method. The ensemble method divides observed unlabeled degradation data into a training dataset and a validation dataset, among which the training dataset trains the model, and the validation dataset is utilized to calculate the weights for each member model. Particularly, we conducted a data augmentation approach using uncertain processes to estimate the initial state parameters of the PF. A weight allocation method using distance measures is presented to assign weights to the prediction results of each member model and perform weighted prediction to obtain the final ensemble prediction results. This paper demonstrates the effectiveness of the proposed ensemble prediction method through two practical examples. Results show that the proposed prediction method has higher accuracy compared to an individual prediction model and mean weight prediction model.

Predicting Lung Cancer Survivability: A Machine Learning Ensemble Method On Seer Data

Ensemble methods are powerful techniques used in machine learning to improve the prediction accuracy of classifier learning systems. In this study, different ensemble learning methods for lung cancer survival prediction were evaluated on the Surveillance, Epidemiology and End Results (SEER) dataset. Data were preprocessed in several steps before applying classification models. The popular ensemble methods Bagging, Adaboost and three classification algorithms, K-Nearest Neighbours, Decision Tree and Neural Networks as base classifiers were evaluated for lung cancer survival prediction. The results empirically showed that ensemble methods are able to evaluate the performance of their base classifiers and they are appropriate methods for analysis of cancer survival.

An integrated approach of ensemble learning methods for stock index prediction using investor sentiments

It has been evidenced by numerous studies that irrational investor sentiment is one of the critical factors leading to dramatic volatility in financial market prices. Therefore, how to effectively predict market prices by information on investor sentiment is a popular but complex topic for researchers, market investors, and financial regulators. In this research, we aim to investigate the effectiveness of stock index prediction using multiple investor sentiment features, and we propose an advanced price trend prediction and trading simulation approach for the Shanghai Stock Exchange index and the Shenzhen Component index by integrating the Boosting, Bagging, and NSGA-II methods. Additionally, the SHAP method is employed as a model interpretation approach to analyze the importance of the sentiment variables and quantify their contributions to the predictions from both local and global perspectives. According to the experimental results, it can be found that the proposed method outperforms the benchmark methods in terms of the hit ratio, accumulated return, and maximum drawdown. It indicates that the proposed method is capable of achieving high accuracy, low risk, and stable profit in price trend prediction and trading simulation of the Chinese stock indexes. Moreover, the SHAP approach incorporated in the proposed method improved the interpretability of the proposed model, which can provide a beneficial reference for market participants to clarify the important sentiment factors and make relative decisions.

SWEP-RF: Accuracy sliding window-based ensemble pruning method for latent sector error prediction in cloud storage computing

Latent sector errors (LSEs) in disk drives cause significant outages, data loss, and unreliability in large-scale cloud storage systems, posing not only technical challenges but also environmental concerns in the context of carbon recycling. Predicting LSEs can help avoid these problems and improve cloud reliability, while also contributing to a more sustainable cloud infrastructure. Ensemble classifiers typically outperform individual classifiers for LSE prediction with high accuracy but can lead to underfitting and incurring additional computational cost, complexity, and time and memory consumption. This research addresses this challenge by proposing a twofold solution: optimizing the ensemble diversity of the resulting Random Forest (RF) classifier through accuracy sliding window-based ensemble pruning (SWEP-RF) and using this pruned ensemble to predict LSEs in cloud storage. By effectively predicting and mitigating LSEs, this approach reduces unnecessary energy consumption and carbon emissions associated with data recovery and reprocessing, aligning with carbon recycling goals. SWEP-RF maximizes its lower margin distribution to adapt the RF prediction performance and produce a strong-performing and effective subensemble, further enhancing the overall energy efficiency of cloud systems. Our approach also reduces ensemble size while maintaining high prediction accuracy, leading to more sustainable resource utilization. We evaluate our algorithm using datasets from Baidu Inc and Backblaze datacenters. Experimental results demonstrate that our approach achieves over 98.6% prediction accuracy, a low false alarm rate (FAR) of 0.003%, and extended meantime to data loss (MTTDL) with lead time in advance (LTA) of up to 383.4 Hrs. and 474.3 Hrs., respectively. SWEP-RF outperforms classical models and current state-of-the-art techniques in prediction accuracy, FAR, MTTDL, processing time, memory consumption, and cloud availability, highlighting its significance in not only enhancing cloud storage reliability but also reducing the carbon footprint of cloud services. Our method is a promising solution for enhancing cloud storage reliability through proactive LSE prediction, while addressing the urgent need for sustainable practices and carbon recycling in cloud computing.

Photovoltaic Power Output Prediction Based on TabNet for Regional Distributed Photovoltaic Stations Group

With the increasing proportion of distributed photovoltaic (DPV) installations in county-level power grids, to improve the centralized operation and maintenance of the stations and to meet the needs of power grid dispatching, the output of the county-level regional DPV stations group needs to be predicted. In this paper, the weather prediction information is used to predict the output based on the model input average strategy. To eliminate the effect of the selected non-optimal training sample collection period on the prediction accuracy, an ensemble prediction method based on the minimum redundancy maximum relevance criterion and TabNet model is carried out. To reduce the influence of weather prediction errors on the power output prediction, a modified model based on error prediction is proposed. The ensemble prediction model is used to predict the day-ahead output, and a combination prediction model based on the proposed ensemble prediction model and the proposed modified model is established to predict the hour-ahead output. The experimental results verify the effectiveness of the proposed models. Compared with the corresponding reference models, the proposed ensemble prediction method reduces the normalized mean absolute errors (nMAEs) and the normalized root mean square errors (nRMSEs) of the day-ahead output prediction results by 2.86% and 5.51%, respectively. The combination prediction model reduces the nMAE and nRMSE of the hour-ahead output prediction results by 3.05% and 3.05%, respectively. Therefore, the prediction accuracy can be improved by the proposed models.

An ensemble learning method for low visibility prediction on freeway using meteorological data

AbstractThe prediction of low visibility is essential for proactive traffic safety management on freeways under fog conditions. However, few studies have developed prediction models for visibility on freeways at a short‐term time interval. This study proposes an ensemble learning approach to develop a short‐term prediction model of low visibility on freeways using meteorological data. Spearman's rank correlation coefficient is used to select meteorological variables related to low visibility. Random forests (RF) and extreme gradient boosting (XGB) are employed to develop visibility prediction models, and back propagation neural network (BPNN) and logistic regression (LR) are used for comparison. The models are evaluated over five prediction time intervals (5, 10, 15, 30, and 60 min). The results indicate that the RF models outperform the other models with precision of 73.9%, recall of 59.8% and F1 score of 0.65. Moreover, the prediction model with a 15‐min time interval shows better performance. With the proposed short‐term prediction of low visibility, it is expected that more crashes could be prevented with more appropriate proactive traffic safety management strategies.

A New Ensemble Prediction Method for Reclaimed Asphalt Pavement (RAP) Mixtures Containing Different Constituents

Fatigue and rutting are two common damage types in asphalt pavements. Reclaimed asphalt pavement (RAP), as a sustainable approach in the pavement industry, deals with the foregoing damage. Fatigue and rutting characteristics of asphalt pavement are generally assessed using laboratory tests, taking a long time and consuming significant amounts of raw material. This study aims to propose a novel approach for predicting fatigue and rutting performance of RAP mixtures. A new ensemble prediction method, named COA-KNN, is introduced by combining the coyote optimization algorithm and K-nearest neighbor to increase the accuracy of fatigue and rutting prediction. In order to evaluate the accuracy, the proposed method was compared against robust prediction methods, including random forest (RF), gradient boosting (GB), decision tree regression (DT), and multiple linear regression (MLR). Afterward, the influence of each variable on the mentioned damages is examined, and the variables are ranked based on their relative influence on the mentioned damages. The results suggest that COA-KNN outperformed other prediction techniques when comparing different performance indicators. Total binder content in asphalt mixes and the PG span of the virgin binder added to the recycled asphalt mixture had the highest relative influence on fatigue and rutting performance, respectively.