PM2.5 Concentration Prediction Model: A CNN–RF Ensemble Framework

Globally, wind energy if properly harnessed, could serve as a source of energy generation in Africa. This study compared the performance of two Machine Learning (ML) algorithms (Linear regression and Random Forest) in predicting wind speed in five major cities in Africa (Yaoundé, Pretoria, Nairobi, Cairo and Abuja). Wind data were collected between January 1, 2000, and December 31, 2022, using the Solar Radiation Data Archive. The data preprocessing was carried out with 80% of the data used for training and 20% for validation. The performance of these ML algorithms was evaluated using Mean Square Error (MSE), Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE) and coefficient of determination (R2). The result shows that Nairobi (3.814795 m/s) closely followed by Cairo (3.606453 m/s) has the highest mean wind speed while Yaoundé (1.090512 m/s) has the lowest. Based on the performance metrics used, the two Machine Learning algorithms were competitive. Still, the Linear Regression (LR) algorithm outperformed the Random Forest Algorithm in predicting wind speed in all the selected major African cities. In Yaoundé (RMSE = 0.3892, MAE= 0.3001, MAPE =0.5030), Pretoria (RMSE=1.2339, MAE=0.9480, MAPE=0.7450) Nairobi (RMSE= 0.4223, MAE =0.6499, MAPE =0.1872), Nairobi (RMSE=0.6499, MAE=0.5171, MAPE =0.1872), Cairo (RMSE =1.0909, MAE =0.8544, MAPE =0.3541) and Abuja (RMSE = 0.70245, MAE =0.5441, MAPE= 0.4515) the Linear regression algorithms was found to outperformed Random Forest Regression. Therefore, the Linear regression algorithm is more reliable in predicting wind speed compared with the Random Forest regression.

<p>Accurate water level (WL) forecasting is important for water resources management and planning purposes in the Great Lakes. The objectives of this research are two-fold.  The first objective is to apply machine learning (ML) (i.e., random forest (RF) and support vector regression (SVR)) and hybrid convolutional neural network(CNN)-long-short term memory (LSTM) deep learning (DL) models for multi-step (i.e., one-, two- and three-monthly step ahead) WL forecasting in the Great Lakes (Michigan and Ontario). The second objective is to integrate the boundary corrected (BC) maximal overlap discrete wavelet transform (MODWT) with SVR, RF, and CNN-LSTM models to improve the performance of the individual models. By employing a BC-wavelet decomposition method, the ‘future data’ issue (i.e., data from the future that is not available), often overlooked in the literature and a major barrier to achieving realistic forecasting performance is overcome. </p><p>For Lakes Michigan and Ontario, 1212 monthly WL (m) records (spanning Jan 1918–Dec 2018) were used to develop the models. For the non-wavelet-based models (SVR, RF, and CNN-LSTM), candidate model inputs included the WL recorded over the previous 12 months.  For the BC-MODWT-based models (BC-MODWT-SVR, BC-MODWT-RF, and BC-MODWT-CNN-LSTM), the lagged input time series were decomposed into BC-wavelet and scaling coefficients by using different mother wavelets (Haar, Daubechies, Symlets, Fejer-Korovkin and Coiflets), filter lengths (from two up to 12) and decomposition levels (from one up to seven).  For each method (SVR, RF, and CNN-LSTM), mother wavelet, and decomposition level a model was generated.  For both wavelet- and non-wavelet-based models, the particle swarm optimization (PSO) method was used to select the most appropriate inputs to include in the proposed multi-step WL forecasting models.</p><p>The datasets were partitioned into calibration and validation subsets. After calibrating the models, various performance evaluation metrics, e.g., coefficient of determination (R<sup>2</sup>), root mean square error (RMSE), mean absolute error (MAE), root mean square percentage error (RMSPE), mean absolute percentage error (MAPE) and the Nash-Sutcliffe efficiency coefficient (NSC) were used to assess model accuracy.</p><p>Of the ML models, the SVR outperformed RF while the DL models outperformed the ML models for each forecast lead time (one-, two-, and three-step(s) ahead). Results from this case study indicate that not all wavelet families and decomposition levels perform equally and in some cases, the wavelet-based models do not improve performance over the non-wavelet-based models. However, the BC-MODWT-CNN-LSTM using suitable mother wavelets (e.g., Haar) outperforms the individual ML and BC-MODWT-ML-based models. More accurate forecasts were obtained for Lake Michigan although the performance in both Great Lakes was accurate. The outcomes of this research indicate that the BC-MODWT-CNN-LSTM model is a promising tool for generating accurate WL forecasts.</p>

Abstract. Cloud fraction as a vital component of meteorological satellite products plays an essential role in environmental monitoring, disaster detection, climate analysis and other research areas. Random forest (RF) and multilayer perceptron (MLP) algorithms were used in this paper to retrieve the cloud fraction of AGRI (Advanced Geosynchronous Radiation Imager) on board the Fengyun-4A (FY-4A) satellite based on its full-disk level-1 radiance observation. Corrections have been made subsequently to the retrieved cloud fraction in areas where solar glint occurs using a correction curve fitted with sunglint angle as weight. The algorithm includes two steps: the cloud detection is conducted firstly for each AGRI field of view to identify whether it is clear sky, partly cloudy or overcast within the observation field. Then, the cloud fraction is retrieved for the scene identified as partly cloudy. The 2B-CLDCLASS-lidar cloud fraction product from the CloudSat and CALIPSO active remote sensing satellite is employed as the truth to assess the accuracy of the retrieval algorithm. Comparison with the operational AGRI level-2 cloud fraction product is also conducted at the same time. The results indicate that both the RF and MLP cloud detection models achieved high accuracy, surpassing that of operational products. However, both algorithms demonstrated weaker discrimination capabilities for partly cloudy conditions compared to clear-sky and overcast situations. Specifically, they tended to misclassify fields of view with low cloud fractions (e.g., cloud fraction = 0.16) as clear sky and those with higher cloud fractions (e.g., cloud fraction = 0.83) as overcast. Between the two models, RF exhibited higher overall accuracy. Both RF and MLP models performed well in cloud fraction retrieval, showing lower mean error (ME), mean absolute error (MAE) and root mean square error (RMSE) compared to operational products. The ME for both RF and MLP cloud fraction retrieval models was close to zero, while RF had slightly lower MAE and RMSE than MLP. During daytime, the high reflectance in sunglint areas led to larger retrieval errors for both RF and MLP algorithms. However, after correction, the retrieval accuracy in these regions improved significantly. At night, the absence of visible light observations from the AGRI instrument resulted in lower classification accuracy compared to daytime, leading to higher cloud fraction retrieval errors during nighttime.

Noninvasive assessment of skin structure using hyperspectral images has been intensively studied in recent years. Due to the high computational cost of the classical methods, such as the inverse Monte Carlo (IMC), much research has been done with the aim of using machine learning (ML) methods to reduce the time required for estimating parameters. This study aims to evaluate the accuracy and the estimation speed of the ML methods for this purpose and compare them to the traditionally used inverse adding-doubling (IAD) algorithm. We trained three models – an artificial neural network (ANN), a 1D convolutional neural network (CNN), and a random forests (RF) model – to predict seven skin parameters. The models were trained on simulated data computed using the adding-doubling algorithm. To improve predictive performance, we introduced a stacked dynamic weighting (SDW) model combining the predictions of all three individually trained models. SDW model was trained by using only a handful of real-world spectra on top of the ANN, CNN and RF models that were trained using simulated data. Models were evaluated based on the estimated parameters’ mean absolute error (MAE), considering the surface inclination angle and comparing skin spectra with spectra fitted by the IAD algorithm. On simulated data, the lowest MAE was achieved by the RF model (0.0030), while the SDW model achieved the lowest MAE on in vivo measured spectra (0.0113). The shortest time to estimate parameters for a single spectrum was 93.05 μs. Results suggest that ML algorithms can produce accurate estimates of human skin optical parameters in near real-time.

As a link for energy transfer between the land and atmosphere in the terrestrial ecosystem, karst vegetation plays an important role. Karst vegetation is not only affected by environmental factors but also by intense human activities. The nonlinear characteristics of vegetation growth are induced by the interaction mechanism of these factors. Previous studies of this relationship were not comprehensive, and it is necessary to further explore it using a suitable method. In this study, we selected climate, human activities, topography, and soil texture as the response factors; a nonlinear relationship model between the karst normalized difference vegetation index (NDVI) and these factors was established by applying a back propagation neural network (BPNN), a radial basis function neural network (RBFNN), the random forest (RF) algorithm, and support vector regression (SVR); and then, the karst NDVI was predicted. The coefficient of determination (R2), mean square error (MSE), root mean square error (RMSE), and mean absolute percentage error (MAPE) of the obtained results were calculated, and the mean R2 values of the BPNN, RBFNN, RF, and SVR models were determined to be 0.77, 0.86, 0.89, and 0.91, respectively. Compared with the BPNN, RBFNN, and RF models, the SVR model had the lowest errors, with mean MSE, RMSE, and MAPE values of 0.001, 0.02, and 2.77, respectively. The results show that the BPNN, RBFNN, RF, and SVR models are within acceptable ranges for karst NDVI prediction, but the overall performance of the SVR model is the best, and it is more suitable for karst vegetation prediction.

To accurately estimate the age of individual tree and to achieve full-cycle sustainable management of natural Larix gmelinii forest in Great Xing'an Mountains of northeastern China, we constructed individual tree age prediction model using stepwise regression and random forest algorithms based on 44 fixed plots data and 280 stan-dard tree cores obtained from the Pangu Forest Farm. We analyzed the influence of stand structure, site conditions, and competition index on the accuracy of model prediction. The model was evaluated by the coefficient of determination (R2), root mean square error (RMSE), and mean absolute error (MAE). The results showed that the random forest model had the highest prediction accuracy when number of decision trees was 1500 and number of node con-tention variables was 8. The random forest model had better accuracy and prediction ability than the stepwise regression model, with R2, RMSE and MAE of 0.5882, 9.9259 a, 8.1155 a. Diameter at breast height was the most important factor affecting age prediction (83.8%), followed by tree height (34.4%), elevation (17.9%), and basal area per hectare (17.5%). The random forest algorithm exhibited better adaptability and modeling effect on constructing a predictive model for individual tree age. This research contributed to improving the accuracy of growth and harvest estimation for L. gmelinii, and could provide a reference for other scientific studies related to tree age estimation in forests.

Accurate prediction of agricultural yield is extremely important to ensure food security and cope with the challenges created by climate change and natural disasters. Forecasting agricultural yield is a challenging task due to the complex nature of variables (fertiliser, rainfall, temperature and others) that affect agricultural production. This study employs six supervised machine learning algorithms: Support Vector Machine (SVM), Decision Tree (DT), Random Forest (RF), Multi-Layer Perceptron (MLP), Recurrent Neural Network (RNN), and Convolutional Neural Network (CNN) to build a predictive model using 49 years of historical data (1973-2021) on paddy, wheat, and maize. Model performance was evaluated using Mean Squared Error (MSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), and Root Mean Squared Error (rMSE). Results show that DT and RF models are the most precise with MSE 1% to 5%, MAE 8% to 21%, followed by SVM and CNN. Key predictors of crop yield include area cultivated, capital expenditure, banking expansion, rainfall, temperature, and fertilizers, while irrigation and road network were less significant. The study recommends that farmers prioritize commercial farming, agricultural equipment, and timely availability of fertilizer for application. The Government of Nepal (GoN) should redirect subsidies towards agricultural mechanization, ensure timely supply of fertilizer, and expand banking services in agricultural areas.

Backbreak is a rock fracture problem that exceeds the limits of the last row of holes in an explosion operation. Excessive backbreak increases operational costs and also poses a threat to mine safety. In this regard, a new hybrid intelligence approach based on random forest (RF) and particle swarm optimization (PSO) is proposed for predicting backbreak with high accuracy to reduce the unsolicited phenomenon induced by backbreak in open-pit blasting. A data set of 234 samples with six input parameters including special drilling (SD), spacing (S), burden (B), hole length (L), stemming (T) and powder factor (PF) and one output parameter backbreak (BB) is set up in this study. Seven input combinations (one with six parameters, six with five parameters) are built to generate the optimal prediction model. The PSO algorithm is integrated with the RF algorithm to find the optimal hyper-parameters of each model and the fitness function, which is the mean absolute error (MAE) of ten cross-validations. The performance capacities of the optimal models are assessed using MAE, root-mean-square error (RMSE), Pearson correlation coefficient (R2) and mean absolute percentage error (MAPE). Findings demonstrated that the PSO–RF model combining L–S–B–T–PF with MAE of 0.0132 and 0.0568, RMSE of 0.0811 and 0.1686, R2 of 0.9990 and 0.9961 and MAPE of 0.0027 and 0.0116 in training and testing phases, respectively, has optimal prediction performance. The optimal PSO–RF models were compared with the classical artificial neural network, RF, genetic programming, support vector machine and convolutional neural network models and show that the PSO–RF model has superiority in predicting backbreak. The Gini index of each input variable has also been calculated in the RF model, which was 31.2 (L), 23.1 (S), 27.4 (B), 36.6 (T), 23.4 (PF) and 16.9 (SD), respectively.

With the development of the Internet of Things (IoT) and virtual reality (VR) technology, the demand for high-precision gesture intelligent analysis of a human–machine interaction module for IoT–VR systems is increasing. Therefore, random forest (RF) and convolution neural network (CNN) algorithms are used in this study to build an intelligent gesture recognition model. The experiments were conducted to test the application performance of the design model. The test results show that the qualification rate of the analytical model designed in this study is significantly higher than that of the comparative model. When the threshold is determined to be 43.26 mm, the analytical qualification rates of the RF-CNN (the method of combining RF with CNN algorithms), faster regions with CNN features (Faster-RCNN), and RF models are 82.41%, 76.10%, and 59.10%, respectively. The calculation time of the RF–CNN model is between the two comparative models. From the test data, it can be observed that the research results have certain significance for improving the accuracy of gesture machine recognition technology in China’s VR Internet of Things (IoT) system.

Soil cadmium (Cd) contamination has attracted a great deal of attention because of its detrimental effects on animals and humans. This study aimed to develop and compare the performances of stepwise linear regression (SLR), classification and regression tree (CART) and random forest (RF) models in the prediction and mapping of the spatial distribution of soil Cd and to identify likely sources of Cd accumulation in Fuyang County, eastern China. Soil Cd data from 276 topsoil (0–20 cm) samples were collected and randomly divided into calibration (222 samples) and validation datasets (54 samples). Auxiliary data, including detailed land use information, soil organic matter, soil pH, and topographic data, were incorporated into the models to simulate the soil Cd concentrations and further identify the main factors influencing soil Cd variation. The predictive models for soil Cd concentration exhibited acceptable overall accuracies (72.22% for SLR, 70.37% for CART, and 75.93% for RF). The SLR model exhibited the largest predicted deviation, with a mean error (ME) of 0.074 mg/kg, a mean absolute error (MAE) of 0.160 mg/kg, and a root mean squared error (RMSE) of 0.274 mg/kg, and the RF model produced the results closest to the observed values, with an ME of 0.002 mg/kg, an MAE of 0.132 mg/kg, and an RMSE of 0.198 mg/kg. The RF model also exhibited the greatest R2 value (0.772). The CART model predictions closely followed, with ME, MAE, RMSE, and R2 values of 0.013 mg/kg, 0.154 mg/kg, 0.230 mg/kg and 0.644, respectively. The three prediction maps generally exhibited similar and realistic spatial patterns of soil Cd contamination. The heavily Cd-affected areas were primarily located in the alluvial valley plain of the Fuchun River and its tributaries because of the dramatic industrialization and urbanization processes that have occurred there. The most important variable for explaining high levels of soil Cd accumulation was the presence of metal smelting industries. The good performance of the RF model was attributable to its ability to handle the non-linear and hierarchical relationships between soil Cd and environmental variables. These results confirm that the RF approach is promising for the prediction and spatial distribution mapping of soil Cd at the regional scale.

Nowadays, machine learning (ML) algorithms are receiving massive attention in most of the engineering application since it has capability in complex systems modeling using historical data. Estimation of power for CMOS VLSI circuit using various circuit attributes is proposed using passive machine learning-based technique. The proposed method uses supervised learning method, which provides a fast and accurate estimation of power without affecting the accuracy of the system. Power estimation using random forest algorithm is relatively new. Accurate estimation of power of CMOS VLSI circuits is estimated by using random forest model which is optimized and tuned by using multiobjective NSGA-II algorithm. It is inferred from the experimental results testing error varies from 1.4% to 6.8% and in terms of and Mean Square Error is 1.46e-06 in random forest method when compared to BPNN. Statistical estimation like coefficient of determination (R) and Root Mean Square Error (RMSE) are done and it is proven that random Forest is best choice for power estimation of CMOS VLSI circuits with high coefficient of determination of 0.99938, and low RMSE of 0.000116.

Hyperspectral Image (HSI) classification is one of the fundamental tasks in the field of remote sensing data analysis. CNN (Convolutional Neural Network) has been proven to be an effective deep learning model, which can extract high-level features directly from the raw data and thereby utilize rich information contained in HSI data. However, labor cost to label enough HIS data for training model is usually expensive, so that it is a strong demand of utilizing limited training data to get a satisfied classification accuracy. In this paper, we put forward a deep cube CNN model – DCNR, which is composed of a cube neighbor HSI pixels strategy, a deep CNN and a random forest classifier. In DCNR model, cubic samples, containing spectral-spatial information, are generated by putting each target pixel and its neighbors together. Then features with high representative ability, extracted by applying a specially designed cube CNN model on each cubic sample, are fed into the random forest classifier for the classification of the target pixel. Results show that DCNR model can achieve classification accuracy of 96.78%, 96.08% and 94.85% on KSC, IP and SA datasets respectively with 20% samples as training set, and 85.03%, 83.45 and 62.17% on KSC, IP and SA datasets respectively with only 1% samples as training set, significantly outperforming random forest and cube CNN models.

Satellites are capable of observing precipitation over large areas and are particularly suitable for estimating precipitation in high mountains and poorly gauged regions. However, the coarse resolution and relatively low accuracy of satellites limit their applications. In this study, a downscaling scheme was developed to obtain precipitation estimates with high resolution and high accuracy in the Heihe watershed. Shannon’s entropy, together with a semi-variogram, was applied to establish the optimal precipitation station network. A combination of the random forest (RF) method and the residual correction approach with the established rain gauge network was applied to downscale monthly precipitation products from Integrated Multi-satellitE Retrievals for Global Precipitation Measurement (IMERG). The results indicated that the RF model showed little improvement in the accuracy of IMERG-based precipitation downscaling. Including residual modification could improve the results of the RF model. The mean absolute error (MAE) and root mean square error (RMSE) values decreased by 19% and 21%, respectively, after residual corrections were added to the RF approach. Moreover, we found that enough rain gauge records are necessary for and remain an important component of tuning model performance. The application of more rain gauges improves the performance of the combined RF and residual modification methods, with the MAE and RMSE values reduced by 8% and 9%, respectively. Residual correction, together with enough precipitation stations, can effectively enhance the quality of the precipitation patterns and magnitudes obtained in the RF downscaling process. The proposed downscaling scheme is an effective tool for increasing the accuracy and spatial resolution of precipitation fields in the Heihe watershed.

Hoax news is a serious problem in this era. Many people are easily led by opinions made by certain people without seeing the truth or looking for existing facts. To overcome this, many researchers have conducted hoax news detection using various algorithms. In some studies, it is said that Random Forest has better performance to overcome this hoax news problem. In other studies, it is also said that CNN has the same level of performance as the Random Forest algorithm. In addition, the problem that is often found is the error in prediction due to improper preprocessing methods. Therefore, in this research, the appropriate preprocessing method is searched by using several preprocessing scenarios for the Convolutional Neural Network (CNN) and Random Forest algorithms. Therefore, in addition to finding the right preprocessing method for each algorithm, a performance comparison is also carried out on the CNN and Random Forest algorithms using a dataset of 4000 news facts from Kompas.com and 4000 hoax news from the turnback.hoax site. the results obtained in this study are random forest has an average model accuracy value of 90% and the CNN algorithm has an average model accuracy value of 60% using the same extraction method, namely TFIDF combined with Ngrams worth one or unigram

Hospital outpatient volume is influenced by a variety of factors, including environmental conditions and healthcare resource availability. Accurate prediction of outpatient demand can significantly enhance operational efficiency and optimize the allocation of medical resources. This study aims to develop a predictive model for daily hospital outpatient volume using the XGBoost algorithm. Meanwhile, the forecasting performance was compared with that of the Seasonal AutoRegressive Integrated Moving Average with exogenous regressors (SARIMAX) and Random Forest (RF) models. The dataset comprises daily climate data (e.g., temperature, precipitation, PM2.5 levels), historical outpatient volume records, and the number of outpatient specialists available each day. The data range involved spans from January 1, 2014, to October 31, 2024. Data preprocessing involved addressing missing values and encoding categorical variables. Model performance was assessed using three metrics: Mean Absolute Error (MAE), Root Mean Squared Error (RMSE) , Mean Absolute Percentage Error (MAPE), and R-squared (R2) metrics. The XGBoost model exhibited superior predictive accuracy compared to both the SARIMAX and RF models, with the lowest MAE, RMSE, MAPE, and the highest R2, successfully capturing key relationships between climate factors, resource availability, and outpatient volume. The number of outpatient specialists, temporal variables (such as year, quarter, month, and weekday), meteorological conditions (average temperature), and air quality (PM2.5) had the most significant impact on the prediction model. This study underscores the potential of machine learning algorithms like XGBoost in effectively predicting hospital outpatient demand. The findings offer valuable insights for hospitals to make proactive adjustments to their resource allocation, thereby improving their service capacity.