Ensemble Regression Research Articles

Re-evaluating winter carbon sink in Southern Ocean by recovering MODIS-Aqua chlorophyll-a product at high solar zenith angles

Satellite ocean color observations are extensively utilized in global carbon sink evaluation. However, the valid coverage of chlorophyll-a concentration (Chla, mg m−3) measurements from these observations is severely limited during autumn and winter in high latitude oceans. The high solar zenith angle (SZA) stands as one of the primary contributors to the reduced quality of Chla products in the high-latitude Southern Ocean during these seasons. This study addresses this challenge by employing a random forest-based regression ensemble (RFRE) method to enhance the quality of Moderate Resolution Imaging Spectroradiometer (MODIS) Chla products affected by high SZA conditions. The RFRE model incorporates the color index (CI), band-ratio index (R), SZA, sensor zenith angle (senz), and Rayleigh-corrected reflectance at 869 nm (Rrc(869)) as predictors. The results indicate that the RFRE model significantly increased the MODIS observed Chla coverage (1.03 to 3.24 times) in high-latitude Southern Ocean regions to the quality of standard Chla products. By applying the recovered Chla to re-evaluate the carbon sink in South Ocean, results showed that the Southern Ocean’s ability to absorb carbon dioxide (CO2) in winter has been underestimated (5.9–18.6 Tg C year−1) in previous assessments. This study underscores the significance of improving the Chla products for a more accurate estimation of winter carbon sink in the Southern Ocean.

Estimation of Cotton SPAD Based on Multi-Source Feature Fusion and Voting Regression Ensemble Learning in Intercropping Pattern of Cotton and Soybean

The accurate estimation of soil plant analytical development (SPAD) values in cotton under various intercropping patterns with soybean is crucial for monitoring cotton growth and determining a suitable intercropping pattern. In this study, we utilized an unmanned aerial vehicle (UAV) to capture visible (RGB) and multispectral (MS) data of cotton at the bud stage, early flowering stage, and full flowering stage in a cotton–soybean intercropping pattern in the Yellow River Delta region of China, and we used SPAD502 Plus and tapeline to collect SPAD and cotton plant height (CH) data of the cotton canopy, respectively. We analyzed the differences in cotton SPAD and CH under different intercropping ratio patterns. It was conducted using Pearson correlation analysis between the RGB features, MS features, and cotton SPAD, then the recursive feature elimination (RFE) method was employed to select image features. Seven feature sets including MS features (five vegetation indices + five texture features), RGB features (five vegetation indices + cotton cover), and CH, as well as combinations of these three types of features with each other, were established. Voting regression (VR) ensemble learning was proposed for estimating cotton SPAD and compared with the performances of three models: random forest regression (RFR), gradient boosting regression (GBR), and support vector regression (SVR). The optimal model was then used to estimate and visualize cotton SPAD under different intercropping patterns. The results were as follows: (1) There was little difference in the mean value of SPAD or CH under different intercropping patterns; a significant positive correlation existed between CH and SPAD throughout the entire growth period. (2) All VR models were optimal when each of the seven feature sets were used as input. When the features set was MS + RGB, the determination coefficient (R2) of the validation set of the VR model was 0.902, the root mean square error (RMSE) was 1.599, and the relative prediction deviation (RPD) was 3.24. (3) When the features set was CH + MS + RGB, the accuracy of the VR model was further improved, compared with the feature set MS + RGB, the R2 and RPD were increased by 1.55% and 8.95%, respectively, and the RMSE was decreased by 7.38%. (4) In the intercropping of cotton and soybean, cotton growing under 4:6 planting patterns was better. The results can provide a reference for the selection of intercropping patterns and the estimation of cotton SPAD.

Investigating the Impact of Maternal Characteristics and Smoking on Birth Weight: An Ensemble Regression Analysis

Investigating the Impact of Maternal Characteristics and Smoking on Birth Weight: An Ensemble Regression Analysis

Fundamental period prediction of infill reinforced concrete structures using an ensemble of regressors

Fundamental period prediction of infill reinforced concrete structures using an ensemble of regressors

Comparative analysis of machine learning models for predicting PM2.5 concentrations using meteorological and chemical indicators

Air pollution significantly impacts human health, causing numerous premature deaths, particularly with the rise in PM2.5 concentrations. Therefore, comparing different machine learning (ML) models for predicting PM2.5 concentration is crucial. This research focuses on six ML models: Linear Regression (LR), Regression Tree (RT), Support Vector Machine (SVM), Ensemble Regression (ERT), Gaussian Process Regression (GPR), and Artificial Neural Networks (ANN). Trained on six years of data (July 2015–December 2021) with optimized hyperparameters, the models consider eight meteorological and chemical indicators as PM2.5 predictors, including temperature, relative humidity, air pressure, O3, SO2, NO2, dew point, and wind speed. Model efficiency is assessed using Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Correlation Coefficient (R), and Coefficient of Determination (R2) values. The models achieve R2 and RMSE values as follows: LR (0.72, 13.52), RT (0.8, 12.156), SVM (0.82, 10.28), ERT (0.81, 11.87), GPR (0.94, 7.65), and ANN (0.99, 2.36). These metrics indicate the superior performance of ANN, with its R2 value approaching 1 and the lowest RMSE compared to other models. The results highlight the effectiveness of ANN, particularly the model with three hidden layers, in predicting PM2.5 concentration. Utilizing ML models for this purpose is crucial for understanding and mitigating the impacts on human health and the environment, with ANN emerging as a promising tool for various investigations.

Machine-learning-based ensemble regression for vehicle-to-vehicle distance estimation using a toe-in style stereo camera

Adjusting the following distance from the front vehicle in highway traffic is important to reduce the risk of collision. Distance estimation is an important research area for advanced driver assistance systems. Therefore, this paper presents a methodology that combines the strengths of several machine learning algorithms using joint decision mechanisms and searches for optimal results for vehicle-to-vehicle distance estimation. The hyperparameter optimization of machine learning models is performed by an iterative algorithm that compares combinations of hyperparameter values. In addition, machine learning algorithms are combined and tested with ensemble learning methods to improve the results obtained. According to the experiments, the ensemble voting regression created by combining extreme gradient boosting, categorical boosting and two multi-layer perceptron models achieves the best result with a mean absolute percentage error value of 1.5444. Considering the comparisons made with other methods, the accuracy of the results obtained with the proposed method is quite high.Code is at: https://github.com/ozgurduran/V2V-VR.

Pig Weight Estimation Method Based on a Framework Combining Mask R-CNN and Ensemble Regression Model.

Using computer vision technology to estimate pig live weight is an important method to realize pig welfare. But there are two key issues that affect pigs' weight estimation: one is the uneven illumination, which leads to unclear contour extraction of pigs, and the other is the bending of the pig body, which leads to incorrect pig body information. For the first one, Mask R-CNN was used to extract the contour of the pig, and the obtained mask image was converted into a binary image from which we were able to obtain a more accurate contour image. For the second one, the body length, hip width and the distance from the camera to the pig back were corrected by XGBoost and actual measured information. Then we analyzed the rationality of the extracted features. Three feature combination strategies were used to predict pig weight. In total, 1505 back images of 39 pigs obtained using Azure kinect DK were used in the numerical experiments. The highest prediction accuracy is XGBoost, with an MAE of 0.389, RMSE of 0.576, MAPE of 0.318% and R2 of 0.995. We also recommend using the Mask R-CNN + RFR method because it has fairly high precision in each strategy. The experimental results show that our proposed method has excellent performance in live weight estimation of pigs.

Learning sparse feature representation for blind quality assessment of night-time images

Capturing Night-Time Images (NTIs) with high-quality is quite challenging for consumer photography and several practical applications. Thus, addressing the quality assessment of night-time images is urgently needed. Since there is no available reference image for such images, Night-Time image Quality Assessment (NTQA) should be done blindly. Although Blind natural Image Quality Assessment (BIQA) has attracted a great deal of attention for a long time, very little work has been done in the field of NTQA. Due to the capturing conditions, NTIs suffer from various complex authentic distortions that make it a challenging field of research. Therefore, previous BIQA methods, do not provide sufficient correlation with subjective scores in the case of NTIs and special methods of NTQA should be developed. In this paper we conduct an unsupervised feature learning method for blind quality assessment of night-time images. The features are the sparse representation over the data-adaptive dictionaries learned on the image exposure and gradient magnitude maps. Having these features, an ensemble regression model trained using least squares gradient boosting scheme predicts high correlated objective scores on the standard datasets.

Ensemble regression based Extra Tree Regressor for hybrid crop yield prediction system

ObjectiveThe worldwide economies are built on agriculture, and plans for food security, resource allocation, and agricultural practices are all heavily influenced by accurate crop production predictions. Predictive models are becoming indispensable tools for predicting crop prospects due to the development of technology based on data. LimitationA significant disadvantage of the ER-ETR for Hybrid Crop Yield Prediction System can involve overfitting, particularly in cases when the dataset is small or the model complexity is not well managed. Inaccurate forecasts based on unreported data and decreased generalization can result from approach. MethodInitially, the dataset is collected from the GitHub and preprocessed using the Standardscaler method. 70 % of the preprocessed data is used as the training set, and the remaining 30 % is used as the testing set. Kernel Principal Component Analysis (KPCA) is employed to extract the feature. The Least Absolute Shrinkage and Selection Operator (LESSO) Regression is used to feature selection.A reliable method for predicting hybrid crop productivity is provided by the suggested ensemble regression that makes use of feature ensemble regression using Extra Tree Regressor (ER-ETR). ResultA simple internet-based programme for immediate forecasting is created using the Python web framework, and the model that has been trained may be used to predict the resulting profitability. Mean absolute error (MAE), mean square error (MSE), root mean square error (RMSE) and R2 were the testing metrics utilized to assess the classification model. With a 95 % accuracy rate, the suggested model is superior to existing models in terms of accuracy in crop production forecasting while still preserving the data's original distribution.Because of the intuitive online interface, stakeholders can forecast immediately and make well-informed decisions on the best use of resources from agriculture. ConclusionThe study creates a hybrid crop yield prediction system using the ER-ETR approach. Agricultural forecasting benefits greatly from its capacity to integrate several models and take advantage of each one's advantages, which improves prediction accuracy and dependability.

Predicting the Arousal and Valence Values of Emotional States Using Learned, Predesigned, and Deep Visual Features.

The cognitive state of a person can be categorized using the circumplex model of emotional states, a continuous model of two dimensions: arousal and valence. The purpose of this research is to select a machine learning model(s) to be integrated into a virtual reality (VR) system that runs cognitive remediation exercises for people with mental health disorders. As such, the prediction of emotional states is essential to customize treatments for those individuals. We exploit the Remote Collaborative and Affective Interactions (RECOLA) database to predict arousal and valence values using machine learning techniques. RECOLA includes audio, video, and physiological recordings of interactions between human participants. To allow learners to focus on the most relevant data, features are extracted from raw data. Such features can be predesigned, learned, or extracted implicitly using deep learners. Our previous work on video recordings focused on predesigned and learned visual features. In this paper, we extend our work onto deep visual features. Our deep visual features are extracted using the MobileNet-v2 convolutional neural network (CNN) that we previously trained on RECOLA's video frames of full/half faces. As the final purpose of our work is to integrate our solution into a practical VR application using head-mounted displays, we experimented with half faces as a proof of concept. The extracted deep features were then used to predict arousal and valence values via optimizable ensemble regression. We also fused the extracted visual features with the predesigned visual features and predicted arousal and valence values using the combined feature set. In an attempt to enhance our prediction performance, we further fused the predictions of the optimizable ensemble model with the predictions of the MobileNet-v2 model. After decision fusion, we achieved a root mean squared error (RMSE) of 0.1140, a Pearson's correlation coefficient (PCC) of 0.8000, and a concordance correlation coefficient (CCC) of 0.7868 on arousal predictions. We achieved an RMSE of 0.0790, a PCC of 0.7904, and a CCC of 0.7645 on valence predictions.

A generalized prediction model for complete performance investigation of pump operated as turbine in microhydro applications using novel approach for improving sustainability

Centrifugal pumps operated in reverse mode (popularly called pump as turbines (PATs)) are well-established as green energy converters, mainly used in microhydro applications due to their lower capital cost. However, the appropriate selection of the pump for reverse-mode application is a critical issue due to the unavailability of its characteristic curve in turbine mode. Therefore, various researchers have proposed models to predict PAT parameters from pump characteristics based on diverse approaches. Still, researchers are striving to develop a prediction model, which can predict the head and flow rates for the PAT with less than ±20 % deviation for both. This work proposed a novel approach to predict the characteristics of PAT with the integration of ensemble regression modelling for Best Efficiency Point (BEP) and from it, by polynomial equations complete characteristics curve. The data from in-house experiments and open literature are used to develop the proposed model excluding the data of four pumps, which are typically used for validation of it. The predicted parameters of PAT from pump characteristics show good agreement for four selected PATs with <10 % maximum deviation. Compared to the literature prediction models, it shows a lower deviation for the same four PATs. A case study based on the field data is carried out to justify the applicability of the proposed model. By using this model selected pump when operated in PAT will produce higher power and generate more revenue compared to literature models, resulting in the economical utilization of resources and improving sustainability. Though the savings are lower, it will be significant for a large number of PAT applications.

Meta-Learning Guided Weight Optimization for Enhanced Solar Radiation Forecasting and Sustainable Energy Management with VotingRegressor

Solar radiation prediction plays a crucial role in renewable energy management, impacting various decision-making processes aimed at optimizing the utilization of solar resources and promoting sustainability. Ensemble regression methods, notably VotingRegressor, have emerged as promising tools for accurate solar radiation forecasting. By integrating predictions from multiple base estimators, ensemble methods have the potential to capture intricate patterns inherent in solar radiation data. However, achieving optimal predictive performance with ensemble methods heavily relies on the careful weighting assigned to each base estimator, presenting a significant challenge. In this study, a novel approach is presented to enhance solar radiation prediction by utilizing meta-learning techniques to optimize the weighting mechanism in the VotingRegressor ensemble. Meta-learning, a subfield of machine learning focusing on learning algorithms across different tasks, provides a systematic framework for learning to learn. This enables models to adapt and generalize more effectively to new datasets and tasks. Our proposed methodology demonstrated significant improvements, with the VotingRegressor with meta-learning techniques achieving an RMSE of 8.7343, an MAE of 5.42145, and an R² of 0.991913. These results mitigate the need for manual weight tuning and improve the adaptability of the VotingRegressor to varying solar radiation conditions, ultimately contributing to the sustainability of renewable energy systems. The methodology involves a comprehensive exploration of meta-learning techniques, encompassing gradient-based optimization, reinforcement learning, and Bayesian optimization.

Prediction of Adolescents' Fluid Intelligence Scores based on Deep Learning with Reconstruction Regularization.

The aim of this study was to develop a predictive model for uncorrected/actual fluid intelligence scores in 9-10 year old children using magnetic resonance T1-weighted imaging. Explore the predictive performance of an autoencoder model based on reconstruction regularization for fluid intelligence in adolescents. We collected actual fluid intelligence scores and T1-weighted MRIs of 11,534 adolescents who completed baseline tasks from ABCD Data Release 3.0. A total of 148 ROIs were selected and 604 features were proposed by FreeSurfer segmentation. The training and testing sets were divided in a ratio of 7:3. To predict fluid intelligence scores, we used AE, MLP and classic machine learning models, and compared their performance on the test set. In addition, we explored their performance across gender subpopulations. Moreover, we evaluated the importance of features using the SHapley Additive Explain method. Results: The proposed model achieves optimal performance on the test set for predicting actual fluid intelligence scores (PCC = 0.209 ± 0.02, MSE = 105.212 ± 2.53). Results show that autoencoders with refactoring regularization are significantly more effective than MLPs and classical machine learning models. In addition, all models performed better on female adolescents than on male adolescents. Further analysis of relevant characteristics in different populations revealed that this may be related to gender differences in underlying fluid intelligence mechanisms. We construct a weak but stable correlation between brain structural features and raw fluid intelligence using autoencoders. Future research may need to explore ensemble regression strategies utilizing multiple machine learning algorithms on multimodal data in order to improve the predictive performance of fluid intelligence based on neuroimaging features.

IPB-MSA&amp;SO4: a daily 0.25° resolution dataset of in situ-produced biogenic methanesulfonic acid and sulfate over the North Atlantic during 1998–2022 based on machine learning

Abstract. Accurate long-term marine-derived biogenic sulfur aerosol concentrations at high spatial and temporal resolutions are critical for a wide range of studies, including climatology, trend analysis, and model evaluation; this information is also imperative for the accurate investigation of the contribution of marine-derived biogenic sulfur aerosol concentrations to the aerosol burden, for the elucidation of their radiative impacts, and to provide boundary conditions for regional models. By applying machine learning algorithms, we constructed the first publicly available daily gridded dataset of in situ-produced biogenic methanesulfonic acid (MSA) and non-sea-salt sulfate (nss-SO4=) concentrations covering the North Atlantic. The dataset is of high spatial resolution (0.25° × 0.25°) and spans 25 years (1998–2022), far exceeding what observations alone could achieve both spatially and temporally. The machine learning models were generated by combining in situ observations of sulfur aerosol data from Mace Head Atmospheric Research Station, located on the west coast of Ireland, and from the North Atlantic Aerosols and Marine Ecosystems Study (NAAMES) cruises in the northwestern Atlantic with the constructed sea-to-air dimethylsulfide flux (FDMS) and ECMWF ERA5 reanalysis datasets. To determine the optimal method for regression, we employed five machine learning model types: support vector machines, decision tree, regression ensemble, Gaussian process regression, and artificial neural networks. A comparison of the mean absolute error (MAE), root-mean-square error (RMSE), and coefficient of determination (R2) revealed that Gaussian process regression (GPR) was the most effective algorithm, outperforming the other models with respect to simulating the biogenic MSA and nss-SO4= concentrations. For predicting daily MSA (nss-SO4=), GPR displayed the highest R2 value of 0.86 (0.72) and the lowest MAE of 0.014 (0.10) µg m−3. GPR partial dependence analysis suggests that the relationships between predictors and MSA and nss-SO4= concentrations are complex rather than linear. Using the GPR algorithm, we produced a high-resolution daily dataset of in situ-produced biogenic MSA and nss-SO4= sea-level concentrations over the North Atlantic, which we named “In-situ Produced Biogenic Methanesulfonic Acid and Sulfate over the North Atlantic” (IPB-MSA&amp;amp;SO4). The obtained IPB-MSA&amp;amp;SO4 data allowed us to analyze the spatiotemporal patterns of MSA and nss-SO4= as well as the ratio between them (MSA:nss-SO4=). A comparison with the existing Copernicus Atmosphere Monitoring Service ECMWF Atmospheric Composition Reanalysis 4 (CAMS-EAC4) reanalysis suggested that our high-resolution dataset reproduces the spatial and temporal patterns of the biogenic sulfur aerosol concentration with high accuracy and has high consistency with independent measurements in the Atlantic Ocean. IPB-MSA&amp;amp;SO4 is publicly available at https://doi.org/10.17632/j8bzd5dvpx.1 (Mansour et al., 2023b).

A Sample of Am and Ap Candidates from LAMOST DR10 (v1.0) Based on the Ensemble Regression Model

Large samples of Am and Ap stars are helpful in studying the interplay between phenomena like atomic diffusion, magnetic fields, and stellar rotation in stellar astrophysics. Existing samples of Am and Ap stars, mostly obtained from spectral data with a signal-to-noise ratio in the g band (S/Ng) greater than 50, can benefit from expansion by exploring spectra with lower S/Ng. Therefore, this paper proposes an ensemble regression model applicable to spectra with a minimum S/Ng of 30. Using the model, we identify 21,361 Am candidates, of which 11,614 are new, and 6182 Ap candidates, of which 4978 are new, from LAMOST DR10. The Am sample size has increased by 60% and the Ap sample size has increased by 180% compared to the previous sample. In terms of effective temperature (T eff), the Am candidates range mainly from 6000 to 8500 K, while the Ap candidates range from 6000 to 11,700 K. The surface gravity ( logg ) distributions for Am and Ap candidates differ in the range of 3.25–4.75 dex. The number of Am candidates increases stepwise, in contrast to the relatively uniform distribution of Ap candidates across the entire surface gravity range. Regarding metallicity ([Fe/H]), Am candidates typically range from −0.75 to 0.38 dex, peaking near 0 dex, while Ap candidates are distributed from −1.38 to 0.38 dex, with a peak near −0.5 dex.

A kind of multi-dot ensemble regression AI detector for lubricating oil additive content based on lambert-beer law

In this work, we propose a Multi-dot Ensemble Regression AI detector (MER) based on the Lambert-Beer law. We pre-trained a model using the infrared spectral data of target additives collected in advance to detect the target additives in unknown oil samples. The algorithm's feasibility was validated by assessing the content of additives in a series of simulated commercial oil samples that were not part of the training set. We established models for three common lubricating oil additives (anti-friction, anti-wear, and antioxidant agents), demonstrating their effectiveness in oil sample detection. Additionally, by comparing with other algorithms, we established the superiority of MER in small-sample learning scenarios.

Seeing beyond the symptoms: biomarkers and brain regions linked to cognitive decline in Alzheimer's disease.

Early Alzheimer's disease (AD) diagnosis remains challenging, necessitating specific biomarkers for timely detection. This study aimed to identify such biomarkers and explore their associations with cognitive decline. A cohort of 1759 individuals across cognitive aging stages, including healthy controls (HC), mild cognitive impairment (MCI), and AD, was examined. Utilizing nine biomarkers from structural MRI (sMRI), diffusion tensor imaging (DTI), and positron emission tomography (PET), predictions were made for Mini-Mental State Examination (MMSE), Clinical Dementia Rating Scale Sum of Boxes (CDRSB), and Alzheimer's Disease Assessment Scale-Cognitive Subscale (ADAS). Biomarkers included four sMRI (e.g., average thickness [ATH]), four DTI (e.g., mean diffusivity [MD]), and one PET Amyloid-β (Aβ) measure. Ensemble regression tree (ERT) technique with bagging and random forest approaches were applied in four groups (HC/MCI, HC/AD, MCI/AD, and HC/MCI/AD). Aβ emerged as a robust predictor of cognitive scores, particularly in late-stage AD. Volumetric measures, notably ATH, consistently correlated with cognitive scores across early and late disease stages. Additionally, ADAS demonstrated links to various neuroimaging biomarkers in all subject groups, highlighting its efficacy in monitoring brain changes throughout disease progression. ERT identified key brain regions associated with cognitive scores, such as the right transverse temporal region for Aβ, left and right entorhinal cortex, left inferior temporal gyrus, and left middle temporal gyrus for ATH, and the left uncinate fasciculus for MD. This study underscores the importance of an interdisciplinary approach in understanding AD mechanisms, offering potential contributions to early biomarker development.

A Novel DWT-ERT-based Fault Location for Distribution Network

Background: A new DWT-ERT-based fault location method is suggested in the IEEE test feeder. Objective: The fault location approach in the distribution network has been proposed in this pa-per that utilizes the discrete wavelet transform (DWT) and ensemble regression tree (ERT). Methods: The fault location methodology has been validated by simulations conducted on an IEEE 13 bus node test feeder. Results: The results show that the suggested solution has low compute burden and memory re-quirements, and is unaffected by system and fault situations. Conclusion: In this study, the fault location approach for the distribution system employing DWT and ERT has been proposed.

A Building Heat Load Prediction Method Driven by a Multi-Component Fusion LSTM Ridge Regression Ensemble Model

Under the background of “double carbon”, building carbon emission reduction is urgent, and improving energy efficiency through short-term building heat load forecasting is an efficient means of building carbon emission reduction. Aiming at the characteristics of the decomposed short-term building heat load data, such as complex trend changes, significant seasonal changes, and randomness, a single-step short-term building heat load prediction method driven by the multi-component fusion LSTM Ridge Regression Ensemble Model (ST-LSTM-RR) is designed and implemented. First, the trend and seasonal components of the heat load are decomposed by the STL seasonal decomposition algorithm, which are fused into the original data to construct three diversified datasets; second, three basic models, namely, the trend LSTM, the seasonal LSTM, and the original LSTM, are trained; and then, the ridge regression model is trained to fuse the predicted values of the three basic models to obtain the final predicted values. Finally, the method of this paper is applied to the heat load prediction of eight groups in a large mountain hotel park, and the root mean square error (RMSE) and mean absolute error (MAE) are used as the evaluation indexes. The experimental results show that the average RMSE and average MAE of the prediction results of the proposed method in this paper are minimized on the eight groups.

Block-wise imputation EM algorithm in multi-source scenario: ADNI case

Alzheimer’s disease is the most common form of dementia and the early detection is essential to prevent its proliferation. Real data available has been of paramount importance in order to achieve progress in the automatic detection despite presenting two major challenges: Multi-source observations containing Magnetic resonance (MRI), Positron emission tomography (PET) and Cerebrospinal fluid data (CSF); and also missing values within all these sources. Most machine learning techniques perform this predictive task by using a single data modality. Nevertheless, the integration of all these sources of evidence could possibly bring a higher performance at different stages of disease progression. The Expectation Maximization (EM) algorithm has been successfully employed to handle missing values, but it is not designed for typical Machine Learning scenarios where an imputation model is created over training data and subsequently applied on a testing set. In this work, we propose EMreg-KNN, a novel supervised and multi-source imputation algorithm. Based on the EM algorithm, EMreg-KNN builds a regression ensemble model for the imputation of future data thus allowing the further utilization of any vector-based Machine Learning method to automatically assess the Alzheimer’s disease diagnosis. Using the ADNI database, the proposed method achieves significant improvements on F1, AUC and Accuracy measures over classical imputation methods for this database using four classification algorithms. Considering these classifiers in four different classification scenarios, our algorithm is experimentally superior in terms of the F measure, in nearly 82% of the cases under evaluation.