A Comparative Study of Using Adaptive Neural Fuzzy Inference System (ANFIS), Gaussian Process Regression (GPR), and SMRGT Models in Flow Coefficient Estimation

Implementing fuzzy SMRGT, ANN, and ANFIS for flow coefficient estimation in Antalya River Basin

This study investigates changes in river flow patterns, in the Hunza Basin, Pakistan, attributed to climate change. Given the anticipated rise in extreme weather events, accurate streamflow predictions are increasingly vital. We assess three machine learning (ML) models - artificial neural network (ANN), recurrent neural network (RNN), and adaptive fuzzy neural inference system (ANFIS) - for streamflow prediction under the Coupled Model Intercomparison Project 6 (CMIP6) Shared Socioeconomic Pathways (SSPs), specifically SSP245 and SSP585. Four key performance indicators, mean square error (MSE), root mean square error (RMSE), mean absolute error (MAE), and coefficient of determination (R2), guide the evaluation. These models employ monthly precipitation, maximum and minimum temperatures as inputs, and discharge as the output, spanning 1985-2014. The ANN model with a 3-10-1 architecture outperforms RNN and ANFIS, displaying lower MSE, RMSE, MAE, and higher R2 values for both training (MSE = 20417, RMSE = 142, MAE = 71, R2 = 0.94) and testing (MSE = 9348, RMSE = 96, MAE = 108, R2 = 0.92) datasets. Subsequently, the superior ANN model predicts streamflow up to 2100 using SSP245 and SSP585 scenarios. These results underscore the potential of ANN models for robust futuristic streamflow estimation, offering valuable insights for water resource management and planning.

Gaussian process regression (GPR) based non-invasive continuous blood pressure prediction method from cuff oscillometric signals

In this paper, a comparative study of adaptive neural fuzzy inference system and support vector machine regression models for the purpose of monthly rainfall prediction has been done. The models were trained, validated and tested using 50 years (1960–2010) of historical climatic data of Varanasi, a district in Uttar Pradesh state of India. The data used as the predictors in these models were monthly relative humidity, atmospheric pressure, average temperature and wind speed of Varanasi. The adaptive neural fuzzy inference system (ANFIS) model used in this study is generated using grid partitioning method and its performance was analyzed. The v-support vector regression (v-SVR) model is also developed for the comparative analysis of the rainfall prediction and the kernel used for this model is Radial Basis Function. The performance criteria of these models were Mean Square Error (MSE), Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Nash–Sutcliffe model efficient coefficient (E) and correlation coefficient R. In this study, it is observed that the performance of adaptive neural fuzzy inference system (ANFIS) model which is optimized using hybrid method has better performance than v-SVR.

Due to the excitability of neurons in the brain, a neurological disorder is produced known as epilepsy. The brain activity of patients suffering from epilepsy is monitored through electroencephalography (EEG). The multivariate nature of features from time domain, frequency domain, complexity and wavelet entropy based, and the statistical features were extracted from healthy and epileptic subjects using the Bonn University database and seizure and non-seizure intervals using the CHB MIT database. The robust machine learning regression methods based on regression, support vector regression (SVR), regression tree (RT), ensemble regression, Gaussian process regression (GPR) were employed for detecting and predicting epileptic seizures. Performance was measured in terms of root mean square error (RMSE), squared error, mean square error (MSE) and mean absolute error (MAE). Moreover, detailed optimization was performed using a RT to predict the selected features from each feature category. A deeper analysis was conducted on features and tree regression methods where optimal RMSE and MSE results were obtained. The best optimal performance was obtained using the ensemble boosted regression tree (BRT) and exponential GPR with an RMSE of 0.47, an MSE (0.22), an R Square (RS) (0.25) and an MAE (0.30) using the Bonn University database and support vector machine (SVM) fine Gaussian with RMSE (0.63634), RS (0.03), MSE (0.40493) and MAE (0.31744); squared exponential GPR and rational quadratic GPR with an RMSE of 0.63841, an RS (0.03), an MSE (0.40757) and an MAE (0.3472) was obtained using the CHB MIT database. A further deeper analysis for the prediction of selected features was performed on an RT to compute the optimal feasible point, observed and estimated function values, function evaluation time, objective function evaluation time and overall elapsed time.

Agriculture, meteorological, and hydrological drought is a natural hazard which affects ecosystems in the central India of Maharashtra state. Due to limited historical data for drought monitoring and forecasting available in the central India of Maharashtra state, implementing machine learning (ML) algorithms could allow for the prediction of future drought events. In this paper, we have focused on the prediction accuracy of meteorological drought in the semi-arid region based on the standardized precipitation index (SPI) using the random forest (RF), random tree (RT), and Gaussian process regression (GPR-PUK kernel) models. A different combination of machine learning models and variables has been performed for the forecasting of metrological drought based on the SPI-6 and 12months. Models were developed using monthly rainfall data for the period of 2000-2019 at two meteorological stations, namely, Karanjali and Gangawdi, each representing a geographical region of Upper Godavari river basin area in the central India of Maharashtra state which frequently experiences droughts. Historical data from the SPI from 2000 to 2013 was processed to train the model into machine learning model, and the rest of the 2014 to 2019-year data were used for testing to forecast the SPI and metrological drought. The mean square error (MSE), root mean square error (RMSE), adjustedR2, Mallows' (Cp), Akaike's (AIC), Schwarz's (SBC), and Amemiya's PC were used to identify the best combination input model and best subregression analysis for both stations of SPI-6 and 12. The correlation coefficient ([Formula: see text]), mean absolute error (MAE), root mean square error (RMSE), relative absolute error (RAE), and root relative squared error (RRSE) were used to perform evaluation for SPI-6 and 12months of both stations with RF, RT, and GPR-PUK kernel models during the training and testing scenarios. The results during testing phase revealed that the RF was found as the best model in forecasting droughts with values of [Formula: see text], MAE, RMSE, RAE (%), and RRSE (%) being 0.856, 0.551, 0.718, 74.778, and 54.019, respectively, for SPI-6 while 0.961, 0.361, 0.538, 34.926, and 28.262, respectively, for SPI-12 scales at Gangawdi station. Further, the respective values of evaluators at Karanjali station were 0.913 and 0.966, 0.541 and 0.386, 0.604 and 0.589, 52.592 and 36.959, and 42.315 and 31.394 for PUK kernel and RT models, respectively, during SPI-6 and SPI-12. Machine learning models are potential drought warning techniques because they take less time, have fewer inputs, and are less sophisticated than dynamic or scientific models.

Since we all have our own set of limitations when it comes to perceiving the world and reasoning profoundly, we are constantly met with uncertainty as a result of a lack of information (lexical impression, incompleteness), as well as specific measurement inaccuracies. It has been found that uncertainty, which shows up as ambiguity, is the root cause of complexity, which is everywhere in the real world. Most of the uncertainty in civil engineering systems comes from the fact that the constraints (parameters) are hard to understand and are described in a vague way. The ambiguity comes from a number of sources, including physical arbitrariness, statistical uncertainty due to using limited information to estimate these characteristics, and model uncertainty due to using overly simplified methods and idealized depictions of actual performances. Thus, it is better to combine fuzzy set theory and fuzzy logic. Fuzzy logic is well-suited to modelling the indeterminacy and ambiguity that results from multiple factors and a lack of data. In order to improve upon a previous predictive model, this paper uses a smart model built on a fuzzy logic system (FLS). Precipitation, temperature, humidity, slope, and land use data were all taken into account as input variables in the fuzzy model. Toprak's original explanation of the simple membership function and fuzzy rules generation technique (SMRGT) was based on the fuzzy-Mamdani methodology and used the flow coefficient as its output. The model's results were compared to available data. The following factors were considered in the comparison: 1) The maximum, minimum, mean, standard deviation, skewness, variation, and correlation coefficients are the seven statistical parameters. 2) Four types of error criteria: Mean Absolute Relative Error (MARE), Mean Square Error (MSE), Mean Absolute Error (MAE), and Root Mean Square Error (RMSE). 3) Scatter diagram.

Chapter 20 - Wavelet decomposition based on Gaussian process regression and multiple linear regression: Monthly reservoir evaporation prediction

Sandstone groundwater salinization modelling using physicochemical variables in Southern Saudi Arabia: Application of novel data intelligent algorithms

Artificial neural network and fuzzy inference technology have been successfully used in various fields in the last decades. In order to combine the advantages of these approaches, the previous researchers came up with a new model named adaptive neural fuzzy inference system (ANFIS), which has been applied to signal processing and the related fields. Hydrological prediction is an important aspect of hydrological services for economy and society. The prediction result not only provides decision support for generation optimization, but also is of great significance to the economical operation of hydropower systems, navigation, flood control and so on. This paper presents the application of adaptive neural fuzzy inference system (ANFIS) on daily runoff time-series prediction at Tongzilin station. To evaluate the performances of the selected ANFIS, comparison was made with the ANN and autoregressive (AR) model. Previous inflows were chosen as input vectors of the three different models. Nash-Sutcliffe efficiency coefficient (NS coefficient), root mean square error (RMSE) and mean absolute relative error (MARE) were chosen to evaluate the performances of our models. The results show that ANFIS not only keep the potential of the ANN whose advantage is to deal with nonlinear problem, but it also eases the model building process and makes the result more stable. As a result, ANFIS can be a recommended daily runoff time-series prediction model.

Machine learning algorithm to forecast ionospheric time delays using Global Navigation satellite system observations

Introduction: Rainfall has the highest variability at time and place scale. Rainfall fluctuation in different geographical areas reveals the necessity of investigating this climate element and suitable models to forecast the rate of precipitation for regional planning. Ardabil province has always faced rainfall fluctuations and shortage of water supply. Precipitation is one of the most important features of the environment. The amount of precipitation over time and in different places is subject to large fluctuations which may be periodical. Studies show that, due to the certain complexities of rainfall, the models which used to predict future values will also need greater accuracy and less error. Among the forecasting models, Arima has more applications and it has replaced with other models. Materials and Methods: In this research, through order 2 Autoregrressive, Winters, and Arima models, monthly rainfalls of Ardabil synoptic station (representing Ardabil province) for a 31-year period (1977-2007) were investigated. To assess the presence or absence of significant changes in mean precipitation of Ardabil synoptic station, rainfall of this station was divided into two periods: 1977-1993 and 1994-2010. T-test was used to statistically examine the difference between the two periods. After adjusting the data, descriptive statistics were applied. In order to model the total monthly precipitation of Ardabil synoptic station, Winters, Autoregressive, and Arima models were used. Among different models, the best options were chosen to predict the time series including the mean absolute deviation (MAD), the mean squared errors (MSE), root mean square errors (RMSE) and mean absolute percentage errors (MAPE). In order to select the best model among the available options under investigation, the predicted value of the deviation of the actual value was utilized for the months of 2006-2010. Results and Discussion: Statistical characteristics of the total monthly precipitation in Ardabil synoptic station indicates that in May, the highest and in August, the lowest monthly total rainfall accounted in this station. Standard deviation of rainfall reached to the lowest level in August and its peak in November. Coefficients of skewness and kurtosis of total rainfall in all seasons, indicates a lack of compliance with normal distribution. From the view of the range of total monthly rainfall, October and August have highest and the lowest tolerance in these parameters, respectively. The results showed that the percentage of the mean absolute error for Arima, Winters and Autoregressive models was 61.82, 148.39 and 81.54 respectively and its R square came to be 88.28, 61.07 and 85.12 respectively. The comparison of the parameters is an indication of the fact that Arima has the highest R square and the lowest mean absolute error of 88.28 and 61.82 respectively than Winters and Autoregressive models. The presence or absence of significant changes in mean precipitation during 1977-1993 and 2010-1994 in Ardabil synoptic station shows that the difference of rainfall is not significant at the 5% error level from statistical point of view. The comparison between the monthly mean rainfall of Ardabil synoptic station in 1994-2010 and 1977-1993 indicates that rainfall has somewhat decreased in the former in recent years. Considering the low average monthly rainfall of Ardabil synoptic station in 1994-2010 compared to 1977-1993 (21.98 versus 26.11 mm), although no statistically significant difference was found in the average rainfall, low rainfall in this station would not be unexpected in the coming years. The comparison of predicted and actual values from 2011 to 2013 in Ardabil synoptic station showed that fitting real data with expected data was relatively acceptable. The observed differences between the actual and predicted values can be related to the influence of rainfalls and many local and dynamical factors of this area. Therefore, it is necessary for climatologists to better explain and predict phenomena besides statistical models and pay more attention to general circulation models (GCM) under different climate conditions. Conclusion: Results of rainfall investigation by order 2 Autoregrressive, Winters, and Arima models showed a descending trend in monthly rainfalls in the coming years across the study location. The results of modeling and analysis of monthly rainfalls in Ardabil synoptic station showed that among these models, Arima was better than the other two because it enjoyed the lowest MAPE and the highest R2. AIC, RMSE and MAD scales of different patterns were calculated and finally, SARIMA(1,1,1)(2,0,1)12 pattern having the lowest AIC, RMSE and MAD was selected as the most appropriate pattern for monthly rainfall forecasting in Ardabil synoptic station.

Analysis of load plays an important role in the operation of modern power systems due to its highly intermittent nature. This manuscript proposes the best approach by comparing results of Hold out, Cross and Re-Substitution validation from hyperparameter tuned Short Term Load Forecasting (STLF). Tree, Neural Network and GPR (Gaussian Process Regression) are three stochastic regression methods used. Each validation procedure is compared with every considered regression method, leading to 9 such combinations. Each combination is analysed with statistical parameters like RMSE (Root Mean Square Error), R Squared, MSE(Mean Square Error), MAE (Mean Absolute Error) and training time. The best approach is further optimised by modifying hyper parameters using Bayesian, Grid Search, and Random Search and most suitable method is proposed. The simulations are performed in Python and MATLAB platforms. The best combination for computation of STLF is K-fold validation with Tree Regression The statistical parameters obtained from the combination are RMSE, R Squared, MSE, MAE and training time of 0.077, 0.88, 0.0059, 0.046, 1.2 respectively. The best method for hyper-parameter tuning is found out to be Grid search with a reduced MSE of 0.0023.

The appraisal of Bitcoin’s price-changing characteristics is extremely difficult because of the nonlinear, nonstationary, effect of multiple uncontrollable factors, and volatile nature. The conventional approaches to machine learning categorization failed to yield accurate results. Additionally, the performance of the presented regression model strategies was evaluated in terms of the mean absolute percentage error (MAPE) between predicted actual values and expected values, as well as the root mean square error (RMSE). These two performance indicators are insufficient to demonstrate the efficacy of Bitcoin price prediction. The aim of this paper is to propose six regression models for Bitcoin price prediction based on historical data from 2014 to 2020. We employ six different regression models. They are CatBoost regressor, gradient boosting regressor, extra tree regressor, AdaBoost regressor, K-nearest neighbor regressor, and the Theil-Sen regressor. Coefficient of determination (R2), mean absolute error (MAE), mean square error (MSE), root mean square error (RMSE), root mean squared logarithmic error (RMSLE), and mean absolute percentage error (MAPE) were used to examine the models’ performance. The experimental results indicated that the extra tree had the highest MAE of 3.1447, RMSLE of 0.553, and MAPE of 0.4783 while the gradient boosting regressor had the highest MSE of 5.4842 and RMSE of 7.4055. Theil-Sen regressor model produced the highest R2 value of 0.4533.KeywordsBitcoin priceCatBoost regressorGradient boosting regressorExtra tree regressorTheil-Sen regressorCoefficient of determinationMean absolute percentage error

In recent years, gallium nitride (GaN) high electron mobility transistors (HEMTs) have come to the forefront of the semiconductor industry because of their exceptional performance in both high‐power and high‐frequency utility. Accurate capacitance modeling is crucial to optimize performance and facilitate energy‐efficient electronic circuit design. In order to reflect the complex nature of the aluminum scandium nitride (AlScN) gate capacitance in GaN HEMTs this study investigates the use of the unique Grünwald‐Letnikov model based on fractional order calculus. The proposed model presents a powerful approach to accurately characterize capacitance since fractional order derivatives allow modeling of non‐integer order systems. Quantitative assessment of the Grünwald‐Letnikov model's accuracy is performed through various error metrics, including mean absolute error (MAE), root mean square error (RMSE), maximum percentage error (MPE), mean absolute percentage error (MAPE), and mean squared error (MSE), by comparing the model's predictions to experimental data. Notably, this model demonstrates remarkable consistency in error metrics, with maximum values of MPE = 0.21%, MAE = 0.05%, MAPE = 0.33%, MSE = 0.01%, and RMSE = 0.09% for the forward scan, and MPE = 0.32%, MAE = 0.04%, MAPE = 0.39%, MSE = 0.01%, and RMSE = 0.08% for the backward scan. These metrics affirm the model's precision in capturing the nuanced capacitance characteristics of GaN HEMT devices. Hence, herein for the first time, the novel Grünwald‐Letnikov model, augmented by fractional order calculus, proves to be a robust tool for accurately characterizing GaN HEMT capacitance. Its ability to seamlessly account for the complexities introduced by using ferroelectric material highlights its potential for advancing semiconductor design and optimizing device performance.