Random Forest Tree Research Articles

Exploring the Impact of Climate Variables on Livestock Anthrax Outbreaks: A Machine Learning Approach

Anthrax, a globally significant disease, poses substantial threats to both livestock and human populations. Timely identification of anthrax outbreaks is paramount to mitigate its impact on animal health, human health, and public safety. This study aims to construct a predictive model for livestock anthrax disease occurrence. By leveraging the potential of advanced Machine-Learning techniques, we projected the likelihood of anthrax outbreaks across India, through incorporating a diverse set of meteorological, and remote sensing parameters. The ultimate goal is to establish a spatial risk map that can serve as an early warning system, aiding in the anticipation and management of future anthrax outbreaks in India's livestock population. Our analysis revealed elevated risk zones for anthrax outbreaks in the southern and north-eastern regions of India, contrasting with medium to low-risk areas in the central parts. Notably, Enhanced Vegetation Index (EVI), Normalized Difference Vegetation Index (NDVI), rainfall, soil moisture, and wind speed emerged as pivotal variables driving the model's predictive accuracy. Among the employed models, the random forest, adaptive boosting, and classification tree analysis approaches showcased superior performance in livestock anthrax risk assessment. The risk map was generated using significant variables by exploiting best fitted models. These findings hold profound implications for policymakers, guiding the targeted deployment of control strategies against anthrax outbreaks. The dynamic risk maps generated through this study enhance public awareness, equipping decision-makers with vital insights for informed action. By spotlighting risk management endeavours, these maps further enhance governance and risk mitigation efforts.

Mapping impervious surface change from remote sensing and GIS data: A case study in Hochiminh city, Vietnam

Impervious surface is artificial surfaces that prevent water from entering the soil. The increase in impervious surface area has led to negative impacts on the urban environment, including an increase in the risk of flooding, a decrease in vegetation cover, and the formation of urban heat islands. This paper presents the results of building a predictive model of impervious surfaces in Hochiminh city from remote sensing and GIS data. Landsat and Sentinel 2 satellite images for the period 2002–2022 are used to classify impervious surfaces and extract input layers about vegetation cover, land surface temperature, combined with GIS data (elevation, slope, aspect, distance to road, distance to hydrology, population density) for modeling and predicting impervious surface changes in future. 03 machine learning algorithms, including Support Vector Machine (SVM), Random Forest (RF), Classification and Regression Trees (CART) and maximum likelihood method are used to classify impervious surfaces from Landsat satellite images, then select the method with the highest accuracy. To predict the future distribution of impervious surface, this study uses Cellular Automata (CA) model and 02 artificial intelligence algorithms (Artificial Neural Network - ANN, Logistic Regression - LR). The results obtained in the study can be effectively used for urban planning, minimizing the impact of the process of increasing the impervious surface on the urban environment

Mesosilicate materials for environmental applications: Adsorption separation analysis via hybrid computational and machine learning tools

In this extensive analysis, we delve into the utilization of advanced hybrid model for molecular separation description via mesoporous materials. The process considered here is adsorption which is utilized for molecular separation in liquid solution. Data for adsorption are obtained in terms of solute concentration by solving mass transfer equations numerically. Then, the data is used for training and validation of several machine learning techniques to establish the hybrid model. For machine learning model, several methods are utilized including three robust tree-based ensemble models, namely Gradient Boosting (GB), Random Forest (RF), and Extra Trees (ET). The methos were utilized to tackle the challenging task of predicting chemical concentrations in a dataset comprising an array of over 19,000 data. These data points are characterized by input variables representing coordinates (x and y), while the output variable encapsulates chemical concentration (C). By subjecting these models to a meticulous optimization process, we achieved substantial enhancements in their predictive capabilities. RF emerged as the standout performer, boasting a remarkable R2 score of 0.99902. ET and GB also demonstrated commendable performance improvements, underscoring the versatility of tree-based ensemble models in delivering precise and reliable predictions for chemical concentrations within intricate and multifaceted datasets.

Stacked ensemble modeling for improved tuberculosis treatment outcome prediction in pediatric cases

SummaryThe promising results of ML (machine learning) methods in various disciplines have led to the frequent use of these methods in health fields such as disease diagnosis, personalized medicine, medical image‐based diagnosis, and predicting the number of deaths and cases in a pandemic. However, a neglected area in the field of healthcare is the lack of study with ML to predict treatment outcomes for tuberculosis (TB) patients, particularly children experiencing failed treatment. This need has become more apparent as the coronavirus pandemic has reversed the gains of health institutions with TB disease, especially in children. Therefore, this article conducted a study using the stacked ensemble ML method to early predict the risk for children experiencing a failed treatment outcome of TB. To fulfill this need and determine the most appropriate technique, a two‐stage methodology was followed in this work. First, predictions were obtained by combining the information gain feature selection (IGFS) approach with a variety of single‐based ML algorithms, including logistic regression (LR), deep belief neural networks (DBN), random forest (RF), and decision tree (DT). Second, the proposed method, which includes a stacked ensemble ML technique, was used. The latter model uses LR as a meta‐learner and the aforementioned single‐based ML algorithms (DBN, LR, RF, and DT). The performance results of ML models used in the two stages were compared, and the proposed model which is the combination of the stack‐based ensemble learning model and the IGFS technique provided better ROC curves, accuracy, precision, and recall results.

Remaining Useful Life Estimation Framework for the Main Bearing of Wind Turbines Operating in Real Time

The prognosis of wind turbine failures in real operating conditions is a significant gap in the academic literature and is essential for achieving viable performance parameters for the operation and maintenance of these machines, especially those located offshore. This paper presents a framework for estimating the remaining useful life (RUL) of the main bearing using regression models fed operational data (temperature, wind speed, and the active power of the network) collected by a supervisory control and data acquisition (SCADA) system. The framework begins with a careful data filtering process, followed by creating a degradation profile based on identifying the behavior of temperature time series. It also uses a cross-validation strategy to mitigate data scarcity and increase model robustness by combining subsets of data from different available turbines. Support vector, gradient boosting, random forest, and extra trees models were created, which, in the tests, showed an average of 20 days in estimating the remaining useful life and presented mean absolute error (MAE) values of 0.047 and mean squared errors (MSE) of 0.012. As its main contributions, this work proposes (i) a robust and effective regression modeling method for estimating RUL based on temperature and (ii) an approach for dealing with a lack of data, a common problem in wind turbine operation. The results demonstrate the potential of using these forecasts to support the decision making of the teams responsible for operating and maintaining wind farms.

Multi-spectral remote sensing and GIS-based analysis for decadal land use land cover changes and future prediction using random forest tree and artificial neural network

The integration of multi-spectral remote sensing and GIS-based analysis is significant for studying land cover changes, providing valuable insights for informed land management and sustainable development. The present study aims to examine land use land cover (LULC) changes of three decades from 1991 to 2021 and predict the future LULC change in Binh Duong province, Vietnam to explore a future research direction on land use change and associated challenges in the study region. Multi-spectral remote sensing data and random forest tree (RFT) were utilized to generate LULC maps. Areal statistics and annual change rate were considered to analyze the categorical land use change detection. Statistical measures such as user's accuracy, producer’s accuracy, kappa coefficient, and confusion matrix were employed to assess the accuracy of LULC classification. To predict future LULC and simulate the spatio-temporal change, we considered previous year LULC maps, independent spatial variables and a combined artificial neural network (ANN) multi-layer perceptron approach. The analysis revealed that there was a huge transition from agricultural lands to residential land with industry and commerce which resulting an expansion of impervious lands and a rapid decline of agricultural land as well as of scrub land and barren lands, and a changeability of forest and plantation, croplands, and waterbodies. Our study revealed that the impervious land has expanded 10 times within 30 years and will increase in the future.

Revolutionizing Cow Welfare Monitoring: A Novel Top-View Perspective with Depth Camera-Based Lameness Classification.

This study innovates livestock health management, utilizing a top-view depth camera for accurate cow lameness detection, classification, and precise segmentation through integration with a 3D depth camera and deep learning, distinguishing it from 2D systems. It underscores the importance of early lameness detection in cattle and focuses on extracting depth data from the cow's body, with a specific emphasis on the back region's maximum value. Precise cow detection and tracking are achieved through the Detectron2 framework and Intersection Over Union (IOU) techniques. Across a three-day testing period, with observations conducted twice daily with varying cow populations (ranging from 56 to 64 cows per day), the study consistently achieves an impressive average detection accuracy of 99.94%. Tracking accuracy remains at 99.92% over the same observation period. Subsequently, the research extracts the cow's depth region using binary mask images derived from detection results and original depth images. Feature extraction generates a feature vector based on maximum height measurements from the cow's backbone area. This feature vector is utilized for classification, evaluating three classifiers: Random Forest (RF), K-Nearest Neighbor (KNN), and Decision Tree (DT). The study highlights the potential of top-view depth video cameras for accurate cow lameness detection and classification, with significant implications for livestock health management.

Comparative Analysis of Diabetic Prediction Using Machine Learning Algorithms

Diabetes mellitus (DM) is a severe worldwide health problem, and its prevalence is quickly growing. It is a spectrum of metabolic illnesses definite by continually increased blood glucose levels. Undiagnosed diabetes can lead to a variety of difficulties, including retinopathy, nephropathy, neuropathy, and other vascular abnormalities. In this context, machine learning (ML) technologies may be mainly useful for early disease identification, diagnosis, and therapy monitoring. The core idea of this study is to detect the strong ML algorithm to forecast it. For this numerous ML algorithms were chosen i.e., support vector machine (SVM), Naïve Bayes (NB), K nearest neighbor (KNN), random forest (RF), logistic regression (LR), and decision tree (DT), according to this work. Two, Pima Indian diabetic (PID) and Germany diabetes datasets were used and the research was implemented using Waikato environment for knowledge analysis (WEKA) 3.8.6 tool. This research discussed performance matrices and error rates of classifiers for both datasets. The outcomes showed that for the PID database (PIDD), SVM works improved with an accuracy of 74% whereas for Germany RF and KNN work improved with 98.7% accuracy. This study can helps healthcare facilities and researchers in understanding the value and application of ML algorithms in predicting diabetes at an initial stage

Research on network intrusion detection based on XGBoost algorithm and multiple machine learning algorithms

Network intrusion detection refers to monitoring and analysing network traffic, system logs and other information to identify abnormal behaviours and attacks in the network, and take timely and appropriate countermeasures to protect the security and stability of the network. This paper investigates the application of seven machine learning methods in network intrusion detection, and evaluates each model by indicators such as precision, accuracy, recall and F1 score. The results show that XGBoost, Random Forest and Decision Tree models have the best prediction results, while Support Vector Machines and Plain Bayes models have poor prediction results. XGBoost, Random Forest and Decision Tree models all belong to the category of integrated learning, which have strong generalisation ability and robustness, can handle high-dimensional and complex datasets and are not prone to overfitting. In addition, they are able to handle non-linear relationships and are suitable for complex classification problems. Catboost and logistic regression models have better prediction results, but their prediction results are also affected by feature engineering. They may under- or over-fit when dealing with high-dimensional, complex datasets. Support Vector Machines and Plain Bayesian Models have poorer prediction results, which is related to their limitations. Support vector machines may experience computational difficulties when dealing with high-dimensional, complex datasets and are weak in dealing with non-linear relationships. The plain Bayesian model assumes that the features are independent of each other, which may not hold true in practical applications, thus affecting the prediction results. The conclusions of this paper are instructive for research and application in the field of network security, and can provide reference and inspiration for research in related fields.

Feature adaptation for landslide susceptibility assessment in “no sample” areas

Given the time-consuming nature of compiling landslide inventories, it is increasingly important to develop transferable landslide susceptibility models that can be applied to regions without existing data. In this study, we propose a feature-based domain adaptation method to improve the transferability of landslide susceptibility models, especially in “no sample” areas. Two typical landslide-prone areas in Fujian province, southeastern China, were chosen as research cases to test the practicality of the transfer effect. Five conventional machine learning algorithms (Support vector machines (SVM), Random Forest (RF), Logistic Regression (LOG), K-nearest neighbor (KNN), and Decision tree (C4.5)) are used to model landslide susceptibility in sampled areas (source domain), and a feature transfer-based landslide susceptibility evaluation model is constructed under coupled feature transfer methods to evaluate the susceptibility of landslide in un-sampled areas (target domain). The results showed that feature transfer can effectively improve the transferability of different machine learning models for cross-regional prediction (The indicators have improved overall by 8.49%), with SVM (increased by 13.68%) and LOG (increased by 10.19%) models showing the most significant improvements. The feature-based domain adaptive method can alleviate the burden of collecting and labeling new data, and effectively improve the assessment performance of machine learning-based landslide susceptibility models in un-sampled areas. This is a new solution for landslide susceptibility assessment in completely “no sample” areas.

Unveiling the Best-fit Model: A Comparative Analysis of Classification Methods in Predicting Student Success

To reduce failure and personalize instruction, educators work to predict student achievement. For this objective, this study compared several categorization techniques. The study investigated techniques employing datasets from Portuguese schools, even though various circumstances make it difficult to gather full data and achieve high accuracy. Upon evaluating the various algorithms, including Random Forest and Decision Trees, the study determined that Random Forest was the most successful model, attaining a 94.55% accuracy rate. This demonstrates how machine learning—more especially, Random Forest—could forecast student achievement. The study opens the door for applying these techniques to early interventions and personalized learning. But more work needs to be done, such as creating publicly accessible educational datasets and investigating different strategies like regression algorithms to manage the nuances of grading systems more effectively.

Changes in smile parameters after surgical-orthodontic treatment for skeletal Class III malocclusion

Background/purposeMany patients pursue attractive smiles, and surgical-orthodontic treatment is a common method. We aimed to analyze the changes in smile parameters after surgical-orthodontic treatment in patients with skeletal Class III pattern and identify factors influencing postoperative outcomes. Materials and methodsThis retrospective study comprised 34 patients with skeletal Class III malocclusion and pure mandible prognathism who received surgical-orthodontic treatment. Hard and soft tissue parameters were assessed through lateral cephalometry, and smile esthetics were evaluated through extraoral photography. Postoperative changes in smile parameters and between-parameter correlations were analyzed. Random forest and decision tree deep learning models were used to identify factors influencing postoperative changes. ResultsSubstantial postoperative changes were observed in the buccal corridor, upper lip height, and smile index. Significant between-sex differences were noted in the upper midline and right chelion. Strong, positive correlations were found between upper lip height and upper lip ratio and between lower lip height and lower lip ratio. By contrast, strong negative correlations were observed between lower teeth exposure and smile index and between interlabial gap and smile index. The highest degrees of postoperative changes were noted in asymmetry-associated smile parameters: lower lip area, interlabial gap, smile index, buccal corridor, and arc ratio. ConclusionConsiderable changes were noted in skeletal, dental, and soft tissue parameters after surgical-orthodontic treatment. However, the changes in smile esthetics were less prominent from the frontal view. As for chin asymmetry, the most to least prominent changes were in the lower lip area, interlabial gap, smile index, buccal corridor, and arc ratio, respectively.

Explainable machine learning methods for predicting water treatment plant features under varying weather conditions

Accurately predicting key features in WWTPs is essential for optimizing plant performance and minimizing operational costs. This study assesses the potential of various machine learning models for predicting the inflow to anoxic sludge reactors. Firstly, it conducts a comprehensive evaluation of diverse machine learning models, including k-Nearest Neighbors (kNN), Random Forest (RF), XGBoost, CatBoost, LightGBM, and Decision Tree Regression (DTR), for predicting the flow into the Anoxic section under various weather conditions (dry, rainy, and stormy). Secondly, the study introduces parsimonious models guided by variable importance from the XGBoost algorithm. Furthermore, the study employs SHAP (SHapley Additive exPlanations) to elucidate model predictions, providing insights into the contribution of each feature. Data from the COST Benchmark Simulation Model (BSM1) is used to verify the investigated models' effectiveness. Each dataset consists of 14 days of influent data at 15-minute intervals, with 80% of the data used for model training. Results show that ensemble learning methods, particularly CatBoost and XGBoost, demonstrate satisfactory predictive results for Anoxic section flow in the presence of increased variability under rainy and stormy conditions. Notably, the CatBoost and XGBoost models achieve average Mean Absolute Percentage Error values of 1.33% and 1.59%, outperforming the other methods.

Predicting immunogenicity of COVID-19 vaccines in hemodialysis patients with renal disease

Individuals who are diagnosed with chronic kidney disease, particularly those receiving maintenance hemodialysis treatment, face a greater likelihood of suffering from severe symptoms and fatality due to COVID-19. This study aimed to explore the optimal vaccination approach for these individuals. The study used data analysis tasks such as data preprocessing, cleaning, and exploration, and machine learning models including linear regression, random forest, XGBoost, gradient boosting, AdaBoost, decision trees, Lasso, and ridge regression were used to construct the predictive model. The study found that the Lasso model performed the best overall in predicting anti-S IgG antibodies levels in response to COVID-19 vaccines for people with kidney failure with MAE of 8.81, RMSE of 19.59, and R2 value of 0.93. The adjusted R2 value for the Lasso model was also 0.93, indicating that the model's ability to explain the variance in the data was not affected by the number of predictors in the model. The Random Forest model best predicted the duration of immunogenicity, with R2 and adjusted R2 values of 0.71 and 0.69, respectively. The ensemble model that includes all eight models, i.e., Ridge, Lasso, Linear Regression, Random Forest, AdaBoost, Gradient Boosting, XGBoost, and Decision Tree, has the best performance with the lowest MAE, the lowest RMSE, the highest R2, and the highest adjusted R2 values of 3.91, 5.00, 0.73, and 0.72, respectively. However, further research is required to validate these models and extend their application to different populations and vaccine types, as well as considering other factors that may affect immune response to COVID-19 vaccines. These findings can be helpful in improving vaccination strategies and promoting public health.

Harnessing AI to unmask Copenhagen's invisible air pollutants: A study on three ultrafine particle metrics

Ultrafine particles (UFPs) are airborne particles with a diameter of less than 100 nm. They are emitted from various sources, such as traffic, combustion, and industrial processes, and can have adverse effects on human health. Long-term mean ambient average particle size (APS) in the UFP range varies over space within cities, with locations near UFP sources having typically smaller APS. Spatial models for lung deposited surface area (LDSA) within urban areas are limited and currently there is no model for APS in any European city. We collected particle number concentration (PNC), LDSA, and APS data over one-year monitoring campaign from May 2021 to May 2022 across 27 locations and estimated annual mean in Copenhagen, Denmark, and obtained additionally annual mean PNC data from 6 state-owned continuous monitors. We developed 94 predictor variables, and machine learning models (random forest and bagged tree) were developed for PNC, LDSA, and APS. The annual mean PNC, LDSA, and APS were, respectively, 5523 pt/cm3, 12.0 μm2/cm3, and 46.1 nm. The final R2 values by random forest (RF) model were 0.93 for PNC, 0.88 for LDSA, and 0.85 for APS. The 10-fold, repeated 10-times cross-validation R2 values were 0.65, 0.67, and 0.60 for PNC, LDSA, and APS, respectively. The root mean square error for final RF models were 296 pt/cm3, 0.48 μm2/cm3, and 1.60 nm for PNC, LDSA, and APS, respectively. Traffic-related variables, such as length of major roads within buffers 100–150 m and distance to streets with various speed limits were amongst the highly-ranked predictors for our models. Overall, our ML models achieved high R2 values and low errors, providing insights into UFP exposure in a European city where average PNC is quite low. These hyperlocal predictions can be used to study health effects of UFPs in the Danish Capital.

Spatiotemporal Landsat-Sentinel-2 satellite imagery-based Hybrid Deep Neural network for paddy crop prediction using Google Earth engine

Rice is one of the predominant food sources to fulfill the dietary requirements of well-being in India. Therefore, accurate and timely paddy crop yield prediction is crucial to ensure the food security of the country. In this direction, the present study proposed a hybrid deep-learning method based on Conv-1D and LSTM layers using the classification-derived phenological with meteorological parameters for paddy crop yield prediction. The paddy crop classification has been conducted using high-resolution (10 m) multispectral imagery based on GPS coordinates collected during the paddy field visits to extract the phenological parameters for input to the prediction model. In this context, the efficiency of Random Forest, Naïve Bayes, SVM, and Gradient Tree boost classifiers was assessed. Furthermore, we have also analyzed the accuracy of Landsat-8, Sentinel-1 GRD, and Sentinel-2 satellite imagery in paddy crop classification based on area estimation. The Statistical Abstract of Haryana was utilized to validate the paddy crop area estimation and yield prediction. The classification outcomes showed that the Random Forest method attained the highest accuracy of 96.6 % compared to other GEE-based classifiers. The proposed Hybrid Deep learning approach achieved an RMSE value of 0.219 t/ha compared to CNN, LSTM, CNN-Bi-LSTM, and Regression techniques for crop yield prediction. The study conclusion highlighted that the sentinel-2 satellite imagery performed well and found that the proposed hybrid approach provided an alternative for paddy crop yield prediction.

Fake News Identification in Urdu Tweets Using Machine Learning Models

There is an increasing number of people who generate and distribute content online, especially via social media platforms, which is primarily responsible for the proliferation of fake information. Fake information can cause controversy and distort people's perspectives, so it needs to be addressed immediately. The goal of this work is to detect false information in Urdu tweets, a difficult task given the language's large user population and particular grammatical difficulties. We offer an all-inclusive machine learning system that reliably classifies tweets in Urdu as legitimate or false. The methodology we use consists of several key steps: preprocessing, which includes normalizing, tokenizing, removing stop words, and stemming to prepare the data for analysis; data collection, which involves compiling and annotating a sizable dataset of Urdu tweets; and feature extraction, which makes use of technique TF-IDF to extract the semantic and syntactic nuances of the language. We investigate various machine learning models, including RNNs and CNNs, and more sophisticated neural networks like SVM, Random Forest, Logistic Regression, Naive Bayes, and Decision Tree to find the most efficient method for resolving this classification problem. The models are put through a rigorous training and assessment process using measures including the F1 score, accuracy, precision, and recall. Furthermore, a thorough examination of their confusion matrices is done. Our study's findings suggest that deep learning models hold much promise for resolving the problem of inaccurate information in Urdu. This opens the door for additional research and the creation of real-time algorithms for spotting false information. The subject of information integrity in Urdu language content is improved by this work, which also sheds light on the applicability of machine learning techniques in many linguistic contexts. Using SVM, Random Forest, Logistic Regression, Naive Bayes, and Decision Tree we achieved accuracies of 74%,91%, 76%, 78%, and 67% respectively. Meanwhile, CNN and RNN are the classifiers with the highest accuracy levels at 91% and 99% respectively. The results demonstrate that the CNN Model achieved 99% highest accuracy in detecting fake news from Urdu Tweets.

An initial investigation into the use of machine learning methods for prediction of carcass component yields in F2 broiler chickens

Context As evaluation of carcass components is costly and time consuming, models for prediction of broiler carcass components are useful. Aims The aim was to investigate the feasibility of machine learning methods in the prediction of carcass components from measurements on live birds during the rearing period. Methods Three machine learning methods, including regression tree, random forest and gradient-boosting trees, were applied to predict carcass yields, and benchmarked against classical linear regression. Two scenarios were defined for prediction. In the first scenario, carcass yields were predicted by live bodyweight, shank length and shank diameter features, recorded at 2, 3 and 4 weeks of age. In the second scenario, predictor features recorded at 5, 6 and 7 weeks of age were used. The two scenarios were reanalysed by including effective single-nucleotide polymorphisms associated with bodyweight, shank length and shank diameter as new predictor features. Key results The correlation coefficient between predicted and observed values for predicting weight of carcass traits ranged from 0.50 for wing to 0.59 for thigh in the first scenario, and from 0.63 for wing to 0.74 for carcass in the second scenario. These predictions for the percentage of carcass components ranged from 0.30 for wing to 0.39 for carcass and breast in the first scenario, and from 0.34 for thigh to 0.43 for carcass in the second scenario when random forest was used. Conclusions Predictive accuracy in the first scenario was lower than in the second scenario for all prediction methods. Including single-nucleotide polymorphisms as predictor features in either scenario did not increase the accuracy of the prediction. Implications In general, random forest had the best performance among machine learning methods, and classical linear regression in two scenarios, suggesting that it may be considered as an alternative to conventional linear models for prediction of carcass traits in broiler chickens.

Ranking building design and operation parameters for residential heating demand forecasting with machine learning

The European Union's Energy Performance in Buildings Directive has made significant strides in enhancing building energy efficiency since its inception in 2002. However, approximately 75% of EU buildings still fall short of energy-efficient standards. Furthermore, there is a growing momentum to extend the concept of nearly zero-energy buildings to entire districts, thereby fostering Net-Zero Energy Districts. This underscores the necessity for large-scale urban building energy modelling to identify and improve underperforming buildings and energy transition planning. Given the increasing interest in black box models for building energy performance, this study aims to identify common input variables in buildings energy demand literature, analyse their influence, and develop a heating demand prediction model using different algorithms: Random Forest, XGBoost, and Extra Trees. Four large datasets generated from white-box simulation in three Spanish cities were used for training and testing the models. Four features consistently stand out as the most important for prediction: shape factor, infiltration rate, south equivalent surface, and internal gains, regardless of the training algorithm or the climatic zone. The multi-location XGBoost algorithm with an optimizer emerged as the best-performing model, with an average Mean Absolute Percentage Error value hovering around 40%. Analysis employing SHapley Additive exPlanation (SHAP) values showcases the model's ability to identify factors that drive higher heating energy demand, alongside its strong predictive performance. This suggests potential integration of the model into programmes to identify key variables to be addressed during building renovation. Additionally, results show the potential of XGBoost-based software's potential for identifying building renovation targets.

Study of the Length of time Window in Emotion Recognition based on EEG Signals

The objective of this research is to present a comparative analysis using various lengths of time windows (TW) during emotion recognition, employing machine learning techniques and the portable wireless sensing device EPOC+. In this study, entropy will be utilized as a feature to evaluate the performance of different classifier models across various TW lengths, based on a dataset of EEG signals extracted from individuals during emotional stimulation. Two types of analyses were conducted: between-subjects and within-subjects. Performance measures such as accuracy, area under the curve, and Cohen's Kappa coefficient were compared among five supervised classifier models: K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Logistic Regression (LR), Random Forest (RF), and Decision Trees (DT). The results indicate that, in both analyses, all five models exhibit higher performance in TW ranging from 2 to 15 seconds, with the 10 seconds TW particularly standing out for between-subjects analysis and the 5-second TW for within-subjects; furthermore, TW exceeding 20 seconds are not recommended. These findings provide valuable guidance for selecting TW in EEG signal analysis when studying emotions.