SVR Model Research Articles

Exploring Hybrid Models For Short-Term Local Weather Forecasting in IoT Environment

This paper explores using and hybridizing simple prediction models to maximize the accuracy of local weather prediction while maintaining low computational effort and the need to process and acquire large volumes of data. A hybrid RF-LSTM model is proposed and evaluated in this research paper for the task of short-term local weather forecasting. The local weather stations are built within an acceptable radius of the measured area and are designed to provide a short period of forecasting - usually within one hour. The lack of local weather data might be problematic for an accurate short-term valuable prediction in sustainable applications like agriculture, transportation, energy management, and daily life. Weather forecasting is not trivial because of the non-linear nature of time series. Thus, traditional forecasting methods cannot predict the weather accurately. The advantage of the ARIMA model lies in forecasting the linear part, while the SVR model indicates the non-linear characteristic of the weather data. Both non-linear and linear approaches can represent the combined model. The hybrid ARIMA-SVR model strengthens the matched points of the ARIMA model and the SVR model in weather forecasting. The LSTM and random forest are both popular algorithms used for regression problems. LSTM is more suitable for tasks involving sequential data with long-term dependencies. Random Forest leverages the wisdom of crowds by combining multiple decision trees, providing robust predictions, and reducing overfitting. Hybrid Random forest-LSTM potentially leverages the robustness and feature importance of Random Forest along with the ability of LSTM to capture sequential dependencies. The comparison results show that the hybrid RF-LSTM model reduces the forecasting errors in metrics of MAE, R-squared, and RMSE. The proposed hybrid model can also capture the actual temperature trend in its prediction performance, which makes it even more relevant for many other possible decision-making steps in sustainable applications. Furthermore, this paper also proposes the design of a weather station based on a real-time edge IoT system. The RF-LSTM leverages the parallelized characteristics of each decision tree in the forest to accelerate the training process and faster inferences. Thus, the hybrid RF-LSTM model offers advantages in terms of faster execution speed and computational efficiency in both PC and Raspberry Pi boards. However, the RF-LSTM consumes the highest peak memory usage due to being a combination of two different models.

Absorption of PV Power Prediction Errors with Headroom Control by Statistical, Machine Learning and Combined Models

Owing to the mass penetration of renewable energy (RE) in Japan, the replacement reserve for feed‐in tariff (RR‐FIT) was introduced in 2021. To promote the further introduction of RE, we attempted to provide reserve power value to the balancing market by controlling the headroom of photovoltaic (PV) power with a PV power plant as a research subject. At the planning stage, the information available is the predicted value, and PV prediction error always occur on the day. Therefore, in this research, we proposed an algorithm to determine the absorption of prediction errors headroom in PV output, thereby maximally reducing negative imbalance. In addition, we developed a statistical model, SVR model, and combined models to calculate and compare the number of imbalance spots and amount of imbalance. Accordingly, the proposed methods can significantly reduce imbalance compared to planning with predicted values, without setting the headroom of prediction errors. © 2023 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC.

A hybrid AHP-PROMETHEE II onshore wind farms multicriteria suitability analysis using kNN and SVM regression models in northeastern Greece

Wind energy presents a high growth potential in the EU as an emission reduction strategy and to achieve the climate neutrality goal by 2050. Wind farms suitability analysis is one of the primary goals in the spatial planning of wind energy developments. This research paper introduces a hybrid spatial multicriteria GIS-based framework that combines Analytic Hierarchy Process (AHP), PROMETHEE II and Machine Learning algorithms to determine and predict the most efficient onshore wind farm locations by generating suitability index mappings. The methodology allows to overcome PROMETHEE II limitations in raster driven suitability analysis, utilizing machine learning regression methods as the k Nearest Neighbor and Support Vector Machines to predict a graduating mapping of suitability index for wind farm locations in northeastern Greece. The best configured models presented a RMSE of 0.0344 and 0.0154 respectively, indicating a quite high predictive performance. Suitability results indicate that 56.10 % of the feasible locations in the Thrace area present a positive outranking character for the kNN model and 56.79 % for the SVR model. The proposed framework, enriched by PROMETHEE II capabilities, assists energy and spatial planners in identifying suitable sites for wind farm siting and enables rational decision making that enhances efficient wind energy investments.

Aboveground Forest Biomass Estimation Using Tent Mapping Atom Search Optimized Backpropagation Neural Network with Landsat 8 and Sentinel-1A Data

Accurate forest biomass estimation serves as the foundation of forest management and holds critical significance for a comprehensive understanding of forest carbon storage and balance. This study aimed to integrate Landsat 8 OLI and Sentinel-1A SAR satellite image data and selected a portion of the Shanxia Experimental Forest in Jiangxi Province as the study area to establish a biomass estimation model by screening influencing factors. Firstly, we extracted spectral information, vegetation indices, principal component features, and texture features within 3 × 3-pixel neighborhoods from Landsat 8 OLI. Moreover, we incorporated Sentinel-1’s VV (vertical transmit–vertical receive) and VH (vertical transmit–horizontal receive) polarizations. We proposed an ensemble AGB (aboveground biomass) model based on a neural network. In addition to the neural network model, namely the tent mapping atom search optimized BP neural network (Tent_ASO_BP) model, partial least squares regression (PLSR), support vector machine (SVR), and random forest (RF) regression prediction techniques were also employed to establish the relationship between multisource remote sensing data and forest biomass. Optical variables (Landsat 8 OLI), SAR variables (Sentinel-1A), and their combinations were input into the four prediction models. The results indicate that Tent_ ASO_ BP model can better estimate forest biomass. Compared to pure optical or single microwave data, the Tent_ASO_BP model with the optimal combination of optical and microwave input features achieved the highest accuracy. Its R2 was 0.74, root mean square error (RMSE) was 11.54 Mg/ha, and mean absolute error (MAE) was 9.06 Mg/ha. Following this, the RF model (R2 = 0.54, RMSE = 21.33 Mg/ha, MAE = 17.35 Mg/ha), SVR (R2 = 0.52, RMSE = 17.66 Mg/ha, MAE = 15.11 Mg/ha), and PLSR (R2 = 0.50, RMSE = 16.52 Mg/ha, MAE = 12.15 Mg/ha) models were employed. In conclusion, the BP neural network model improved by tent mapping atom search optimization algorithm significantly enhanced the accuracy of AGB estimation in biomass studies. This will provide a new avenue for large-scale forest resource surveys.

Predicting Binding Affinity Between MHC-I Receptor and Peptides Based on Molecular Docking and Protein-peptide Interaction Interface Characteristics

Background:Predicting protein-peptide binding affinity is one of the leading research subjects in peptide drug design and repositioning. In previous studies, models constructed by researchers just used features of peptide structures. These features had limited information and could not describe the proteinpeptide interaction mode. This made models and predicted results lack interpretability in pharmacy and biology, which led to the protein-peptide interaction mode not being reflected. Therefore, it was of little significance for the design of peptide drugs.Objective:Considering the protein-peptide interaction mode, we extracted protein-peptide interaction interface characteristics and built machine learning models to improve the performance and enhance the interpretability of models.Methods:Taking MHC-I protein and its binding peptides as the research object, protein-peptide complexes were obtained by molecular docking, and 94 protein-peptide interaction interface characteristics were calculated. Then ten important features were selected using recursive feature elimination to construct SVR, RF, and MLP models to predict protein-peptide binding affinity.Results:The MAE of the SVR, RF and MLP models constructed using protein-peptide interaction interface characteristics are 0.2279, 0.2939 and 0.2041, their MSE are 0.1289, 0.1308 and 0.0780, and their R2 reached 0.8711, 0.8692 and 0.9220, respectively.Conclusion:The model constructed using protein-peptide interaction interface characteristics showed better prediction results. The key features for predicting protein-peptide binding affinity are the bSASA of negatively charged species, hydrogen bond acceptor, hydrophobic group, planarity, and aromatic ring.

Using Different Machine Learning Algorithms to Predict the Prices of Flight Tickets

The rising prices of flight tickets and the lack of transparency in the dynamic pricing strategies of airlines have caused many consumers to wonder, what factors actually determine these prices. In order to investigate this question, a large dataset of flight ticket bookings that includes the most price-defining variables was acquired. This data was preprocessed using discretization, normalization, and principal component analysis. This preprocessed data was then used to train 5 different Machine Learning algorithms: Linear Regression, DecisionTree, Ridge Regression, RandomForest, and SVR. The training of the RandomForest and SVR models was not possible due to runtime errors, however, the other models trained as expected. All models performed well, with the Linear Regression and Ridge Regression performing identically. Overall, the DecisionTree model performed the best at predicting the prices of flights, and by adjusting hyperparameters the performance could be further increased. The investigation could be continued by using a larger dataset to investigate how the model performs with more variables and under broader conditions. Additionally, the model could be reappropriated to make a user-friendly flight price prediction tool that helps consumers with their purchasing decisions.

머신러닝을 이용한 기후변화에 따른 천궁 생리 활성 성분 예측 모델 연구

Climate change has emerged as a global problem, with frequent temperature increases, droughts, and floods, and it is predicted that it will have a great impact on the characteristics and productivity of crops. Cnidium officinale is used not only as traditionally used herbal medicines, but also as various industrial raw materials such as health functional foods, natural medicines, and living materials, but productivity is decreasing due to threats such as continuous crop damage and climate change. Therefore, this paper proposes a model that can predict the physiologically active ingredient index according to the climate change scenario of Cnidium officinale, a representative medicinal crop vulnerable to climate change. In this paper, data was first augmented using the CTGAN algorithm to solve the problem of data imbalance in the collection of environment information, physiological reactions, and physiological active ingredient information. Column Shape and Column Pair Trends were used to measure augmented data quality, and overall quality of 88% was achieved on average. In addition, five models RF, SVR, XGBoost, AdaBoost, and LightBGM were used to predict phenol and flavonoid content by dividing them into ground and underground using augmented data. As a result of model evaluation, the XGBoost model showed the best performance in predicting the physiological active ingredients of the sacrum, and it was confirmed to be about twice as accurate as the SVR model.

Fatigue life prediction of AA2524 thin plate strengthened using compound laser heating and laser shot peening method

AA2524 is a aluminum alloy based on damage tolerance design, and it is mainly used for the manufacturing of aircraft skins. Aircraft skins often occur fatigue failure due to harsh working conditions. To improve the fatigue resistance of aircraft skins and to accurately predict the fatigue life of AA2524 compact tension (CT) samples. The compound strengthening method of lsear heating (LH) and lsear shot peening (LSP) is proposed to improve the fatigue life of AA2524 in this paper. Considering the effect of stress ratio (R), residual stress (RS), stress intensity factor range (ΔK), and maximum stress intensity factor (Kmax) on the fatigue crack growth rate (FCGR), the fatigue life of CT samples was predicted applying artificial neural networks, support vector regression models. The results show that laser heating and laser shot peening can obviously improve the fatigue life by 30.703% at R = 0.1 and 4.701% at R = 0.5, compared with base materials. The neural networks have better fatigue life prediction ability, and the fit coefficients for the neural networks predicting fatigue life reach 0.97, which is better than that of Walker model (0.91). The prediction value for ANN model is 75,462 cycles, the magnitude for SVR model is 78,234 cycles, and the experimental value is 74,204 cycles. The deviation are 1.7% and 5.43%, respectively. Meanwhile, the effect of R on the FCGR of AA2524 is closely associated with the crack closure effect.

Digital Twin-Driven Framework for TBM Performance Prediction, Visualization, and Monitoring through Machine Learning

The rapid development in underground infrastructure is encouraging faster and more modern ways, such as TBM tunneling, to meet the needs of the world. However, tunneling activities generate complex and heterogeneous data, which makes it difficult to visualize the performance of a project. Advancements in information technology, such as digital twins and machine learning, provide platforms for digital demonstration, visualization, and system performance monitoring of such data. Therefore, this study proposes a digital twin-driven framework for TBM performance prediction through machine learning, visualization, and monitoring. This novel approach integrates machine learning and real-time performance data to predict, visualize, and monitor the status of the tunnel construction progress. A digital twin virtual model of TBM was constructed based on TBM design parameters, the input parameter, boring energy, RPM, torque, thrust force, speed, gripper pressure, total revolution, and Q-value provided to SVR and ANN models to predict the TBM AR and PR, and TBM daily progress was visualized continuously. The predictive performance indices R2 (0.97) and RMSE (0.011) were estimated for AR prediction, showing the accuracy of the proposed model. To demonstrate the proposed framework, this study shows the its effectiveness. By implementing this framework, stakeholders can minimize the risk associated with the cost and schedule of a tunneling project by simultaneously visualizing and monitoring the performance of TBMs through digital twin and machine learning algorithms.

Forecast stock price based on GRA-LoGo model of information filtering networks

The considerable fluctuation of the stock market caused by COVID-19 tends to have a tremendous and long-lasting adverse impact on the economy. In this work, we propose a novel methodology to investigate this impact on the Chinese medical stock market. We examine changes in the stock network structure using the Triangulated Maximally Filtered Graph (TMFG), which is computationally faster and more adaptable to enormous datasets. Additionally, we develop the LoGo model, which combines a local-global approach in its construction, to predict the stock prices of the Chinese medical stock market. In addition to traditional predictors, we incorporate daily new infected numbers as an additional predictor to reflect the impact of COVID-19. We select data from the 2019-2020 period and divide it into two datasets: one for the period during COVID-19 and another for the period before COVID-19. Firstly, we compute the grey correlation coefficients between stocks instead of standard correlation coefficients. We use these coefficients to build the TMFG, enabling us to identify which stocks played the leading roles. Subsequently, we choose six stocks to build the price prediction models. Compared with the LSTM and SVR models, the LoGo models demonstrates higher accuracy, achieving an average accuracy of 71.67 percent. Furthermore, the execution time of the Logo models is 200 times faster than that of the SVR models and 50 times faster than that of the LSTM models.

COOT–SVR model: estimation of the capillary water absorption of building stones

COOT–SVR model: estimation of the capillary water absorption of building stones

Forecasting crude oil futures price using machine learning methods: Evidence from China

Crude oil is an indispensable energy resource. With the establishment of the local crude oil futures market in China, providing accurate forecasts for crude oil futures price is urgent. To cope with this challenge, this paper introduces artificial intelligence methods to evaluate the optimal forecasting strategy for China's crude oil futures price. We use machine learning to process historical information, volatility and non-linear features. Using daily data from March 26, 2018 to February 28, 2023, we estimate the forecasting effects of RNN, LSTM, GRU, SVR, MLP, CNN and BP models on China crude oil futures, respectively. With a series of evaluation tests, we demonstrate that the GRU model outperforms other models in terms of forecast accuracy and performance for China's crude oil futures price. Taking multiple influencing factors into account, the forecasting accuracy of proposed models is improved by including influential factors. Therefore, these findings effectively explore the forecasting of China's crude oil futures prices, contributing to the improvement of the emerging crude oil futures market and the management of energy price risks.

A LATIN AMERICAN MARKET ASSET VOLATILITY ANALYSIS: A COMPARISON OF GARCH MODEL, ARTIFICIAL NEURAL NETWORKS AND SUPPORT VECTOR REGRESSION

The objective of this research was to compare the effectiveness of the GARCH method with machine learning techniques in predicting asset volatility in the main Latin American markets. The daily squared return was utilized as a volatility indicator, and the accuracy of the predictions was assessed using root mean square error (RMSE) and mean absolute error (MAE) metrics. The findings consistently demonstrated that the linear SVR-GARCH models outperformed other approaches, exhibiting the lowest MAE and MSE values across various assets in the test sample. Specifically, the SVRGARCH RBF model achieved the most accurate results for the IPC asset. It was observed that GARCH models tended to produce higher volatility forecasts during periods of heightened volatility due to their responsiveness to significant past changes. Consequently, this led to larger squared prediction errors for GARCH models compared to SVR models. This suggests that incorporating machine learning techniques can provide improved volatility forecasting capabilities compared to the traditional GARCH models.

Retracted: Improved CEEMDAN, GA, and SVR Model for Oil Price Forecasting.

[This retracts the article DOI: 10.1155/2022/3741370.].

Very High Cycle Fatigue Life Prediction of SLM AlSi10Mg Based on CDM and SVR Models

A novel fatigue evolution model considering the effect of defect size and additive manufacturing building direction based on the theories of continuum damage mechanics and its numerical implementation in ABAQUS is proposed in this paper. First, the constitutive model, fatigue damage evolution model and their parameter calibration methods are presented. Second, using the ABAQUS platform, the proposed model is implemented with user-defined subroutines. After that, based on the proposed model and its numerical implementation, the fatigue life of additively manufactured AlSi10Mg is predicted and its applicability is verified through experimental results. Finally, a support vector regression model is established to predict the fatigue life, and its results are compared to those of the numerical finite element method. The results show that the support vector regression model makes better predictions than the finite element method.

Hybrid machine learning model for hourly ozone concentrations prediction and exposure risk assessment

With the development and improvement of optimization algorithms, machine learning models have become more important in air quality prediction. These models can be used to address problems associated with atmospheric environmental pollutants like ozone and facilitate the development of appropriate control policies. Firstly, the data sets of meteorological and pollution in Nanjing and four nearby cities from 2018 to 2021 were divided into four sections to build mixed models. The extreme gradient boosting (XGBoost) algorithm was improved by the permutation importance method (PIM) and Pearson correlation coefficient to acquire the most important features of the seasonal model and analysis of the effect on ozone prediction. The results reflect that the NO2, PM2.5, and PM10 parameters have an important influence, accounting for an average of 23% of prediction, and the average influence of the meteorological parameters of the solar radiation exceeds 11%. Secondly, the particle swarm optimization (PSO), grey wolf optimizer (GWO), and ant lion optimizer (ALO) algorithms were used to optimize the SVR model parameters to predict hourly ozone concentration in January, April, July, and September 2022. The R2 values are 0.95, 0.92, 0.85, and 0.84, respectively, and the RMSE values are 5.2, 12.4, 18.0 and 16.6 μg/m3, respectively. Finally, the impact of hourly ozone concentration on population health was studied by a risk assessment of the population density-weighted pollution exposure. The result shows that the influence on humans gradually decreased yearly from 2018 to 2021, the average variation from 2.6% to −1.1%, and health risks during the evening rush hours are still prevalent and require further attention.

Research on Yield Prediction Technology for Aerospace Engine Production Lines Based on Convolutional Neural Networks-Improved Support Vector Regression

Improving the prediction accuracy of aerospace engine production line yields is of significant importance for enhancing production efficiency and optimizing production scheduling in enterprises. To address this, a novel method called Convolutional Neural Networks-Improved Support Vector Regression (CNN-ISVR) has been proposed for predicting the production line yield of aerospace engines. The method first divides the factors influencing production line yield into production cycle and real-time status information of the production line and then analyzes the key feature factors. To solve the problem of poor prediction performance in traditional SVR models due to the subjective selection of kernel function parameters, an improved SVR model is presented. This approach combines the elite strategy genetic algorithm with the hyperparameter optimization method based on grid search and cross-validation to obtain the best penalty factor and kernel function width of the Radial Basis Function (RBF) kernel function. The extracted features of production data are then used for prediction by inputting them into the improved support vector regression model, based on a shallow CNN without dimensionality reduction. Finally, the prediction accuracy of the CNN-ISVR model is evaluated using the three commonly used evaluation metrics: Mean Absolute Percentage Error (MAPE), Root Mean Square Error (RMSE) and coefficient of determination (R2). The model’s prediction results are then compared to those of other models. The CNN-ISVR hybrid model is shown to outperform other models in terms of prediction accuracy and generalization ability, demonstrating the effectiveness of the proposed model.

Trend Forecasting of Public Concern about Low Carbon Based on Comprehensive Baidu Index and Its Relationship with CO2 Emissions: The Case of China

Climate change is harmful to ecosystems and public health, so the concern about climate change has been aroused worldwide. Studies indicated that greenhouse gas emission with CO2 as the main component is an important factor for climate change. Countries worldwide are on the same page that low-carbon development is an effective way to combat climate change. Enhancing public concern about low-carbon development and climate change has a positive effect on universal participation in carbon emission reduction. Therefore, it is significant to study the trend of public concern about low carbon and its relationship with CO2 emissions. Currently, no related studies are available, so this research explores the relationship between the public concern about low carbon and CO2 emissions of China, as well as the respective trends of each. Based on the daily data of Baidu-related keyword searches and CO2 emission, this research proposes the GMM-CEEMD-SGIA-LSTM hybrid model. The GMM is utilized to construct a comprehensive Baidu index (CBI) to reflect public concern about low carbon by clustering keywords search data. CEEMD and SGIA are applied to reconstruct sequences for analyzing the relationship between CBI and CO2 emissions. Then LSTM is utilized to forecast CBI. The reconstructed sequences show that there is a strong correlation between CBI and CO2 emissions. It is also found that CBI affects CO2 emissions, with varying effect lag times for different periods. Compared to LSTM, RF, SVR, and RNN models, the proposed model is reliable for forecasting public concern with a 46.78% decrease in MAPE. The prediction results indicate that public concern about low carbon shows a fluctuating upward trend from January 2023 to January 2025. This research could improve understanding of the relationship between public concern about low carbon and CO2 emissions to better address climate change.

A novel GA-LSTM-based prediction method of ship energy usage based on the characteristics analysis of operational data

Optimization of ship energy efficiency is an efficient measure to decrease fuel usage and emissions in the shipping industry. The accurate prediction model of ship energy usage is the basis to achieve optimization of ship energy efficiency. This study investigates the sequential properties of the actual voyage data from a VLOC. On this basis, a model for predicting ship energy consumption is established by adopting a LSTM neural network that has better prediction performance for sequential datasets. To further enhance the performance of the established LSTM-based model, the network structures and hyperparameters are optimized by using Genetic Algorithm. Lastly, the application analysis is conducted to validate the established GA-LSTM-based model for ship fuel usage prediction. The established model for ship energy usage shows a significant improvement in prediction accuracy, compared to the original LSTM-based model. Meanwhile, the developed prediction model is more accurate than the existing BP, SVR, and ARIMA-based energy consumption models. The prediction errors for the ship's operational energy efficiency adopting the established GA-LSTM-based model can reach as low as 0.29%. Therefore, the established model can effectively predict the ship fuel usage under different conditions, which is essential for the optimization and improvement of ship energy efficiency.

Short-term streamflow modeling using data-intelligence evolutionary machine learning models

Accurate streamflow prediction is essential for efficient water resources management. Machine learning (ML) models are the tools to meet this need. This paper presents a comparative research study focusing on hybridizing ML models with bioinspired optimization algorithms (BOA) for short-term multistep streamflow forecasting. Specifically, we focus on applying XGB, MARS, ELM, EN, and SVR models and various BOA, including PSO, GA, and DE, for selecting model parameters. The performances of the resulting hybrid models are compared using performance statistics, graphical analysis, and hypothesis testing. The results show that the hybridization of BOA with ML models demonstrates significant potential as a data-driven approach for short-term multistep streamflow forecasting. The PSO algorithm proved superior to the DE and GA algorithms in determining the optimal hyperparameters of ML models for each step of the considered time horizon. When applied with all BOA, the XGB model outperformed the others (SVR, MARS, ELM, and EN), best predicting the different steps ahead. XGB integrated with PSO emerged as the superior model, according to the considered performance measures and the results of the statistical tests. The proposed XGB hybrid model is a superior alternative to the current daily flow forecast, crucial for water resources planning and management.