Mean Average Percentage Error Research Articles

Online prediction of effluent COD in the anaerobic wastewater treatment system based on PCA-LSSVM algorithm.

Since anaerobic wastewater treatment is a nonlinear and complex biochemical process, reasonable monitoring and control are needed to keep it operating stably and efficiently. In this paper, a least-square support-vector machine (LS-SVM) was employed to construct models for the prediction of effluent chemical oxygen demand (COD) in an anaerobic wastewater treatment system. The result revealed that the performance of the steady-state model based on LS-SVM for predicting effluent COD was acceptable, with the maximum relative error (RE) of 11.45%, the mean average percentage error (MAPE) of 0.79% and the root mean square error (RMSE) of 3.08 when training, and the performance fell slightly when testing. Even though, the correlation coefficient value (R) between the predicted value and the actual value of 0.9752 could be achieved, which means this model can predict the variation of effluent COD in general. The dynamic-state models under three kinds of shock loads, which were concentration, hydraulic, and bicarbonate buffer absent, showed good forecasting performance, the correlation coefficient values (R) all excelled 0.99. Among these three shocks, the dynamic LS-SVM model under bicarbonate buffer absent shock achieved the optimal performance and followed by the dynamic-state model under hydraulic shock. This paper provides a meaningful reference to improve the monitoring level of the anaerobic wastewater treatment process.

Planning for the future: Modelling daily emergency department presentations in an Australian capital city.

To describe and model a decade of ED presentations in metropolitan Melbourne, Australia, from July 2000 to June 2010 and to validate the model of ED presentations by testing the model's performance in forecasting the subsequent 2 year period of daily presentations, from July 2010 to June 2012. Retrospective analyses of prospectively collected data sourced from the Victorian Emergency Minimum Dataset were performed and included 13 public hospitals. Time series modelling involved unobserved components modelling and forward selection of variables using incidence rate ratios. Forecasting with the model and validation were performed using the 2 year period up to June 2012. Model performance was calculated using the mean average percentage error. A total of 7 031 242 patient presentations occurred to the sample metropolitan EDs in the 12 year study period. An absolute increase in mean daily ED presentations of 81.3% was observed. Presentations increased on Sunday and Monday incidence rate ratio of 1.10 (95% CI 1.08-1.11, P < 0.05). No monthly or seasonal pattern was evident. Public holidays were associated with increased presentations, incidence rate ratio of 1.11 (95% CI 1.08-1.15, P < 0.05). The model with the best goodness-of-fit and Wald χ2 value included Sunday-Monday (vs Tuesday-Saturday), public holidays, the trend of gradual increase over time and a stochastic (random white noise) cycle. The mean average percentage error for the 2 year forecast period was 3.6%. We have produced and validated a model for predicting daily ED presentations across a major city. Even though ED presentations are multifactorial, city-wide daily presentations are predictable and explained by a small number of variables. The model will have implications for future health planning.

Implementation of Artificial Neural Network to Predict S&P 500 Stock Closing Price

Artificial Neural Network (ANN) is a learning method that can be used for prediction and classification. In this study, ANN back-propagation is implemented to predict the closing price of the S&P 500 stock exchange using historical data. Historical data consisting of five variables, namely open, high, low, close, and volume. The historical data are taken from finance.yahoo.com which stores historical data of daily stock prices up to 65 years earlier. The data is designed based on daily prediction scenarios of the closing stock price of the S&P 500 stock exchange. Using the scenario an application prototype using R and Java software has been successfully built. The prototype is dynamic during select the data set (10, 50, 100, 500 past data) to get the best stock closing prediction for the next day. The selection of the best stock closing prediction uses the MAPE (Mean Average Percentage Error) criterion which is in contrast to the prediction accuracy level. The smaller the MAPE value the better the predicted result. Based on the implementation results, the average MAPE for daily forecast for 1 month in April 2017 was 0.2307 indicating that the average daily prediction accuracy rate for 1 month was 99.77%. Our prototype application can make prediction automatically every day, every time when connected to the internet.

Computational and experimental study for the desalination of petrochemical industrial effluents using direct contact membrane distillation

The petrochemical, mining and power industries have reacted to the recent South African water crisis by focussing on improved brine treatment for water and salt recovery with the aim of achieving zero liquid effluent discharge. The purpose of this novel study was to compare experimentally obtained results from the treatment of synthetic NaCl solutions and petrochemical industrial brines such as spent ion exchange regenerant brines and reverse osmosis (RO) brines to the classical well-known Knudsen diffusion, molecular diffusion and transition predictive models. The predictive models were numerically solved using a developed mathematical algorithm that was coded using MATLAB® software. The impact of experimentally varying the inlet feed temperature on process performance of the system is presented here and compared to simulated results. It was found that there was good agreement between the experimentally obtained results, for both the synthetic NaCl solution and the industrial brines. The mean average percentage error (MAPE) was found to be 7.9% for the synthetic NaCl solutions when compared to the Knudsen model. The Knudsen/molecular diffusion transition theoretical model best predicted the performance of the membrane for the industrial spent ion exchange regenerant brine with a mean absolute percentage error (MAPE) of 13.3%. The Knudsen model best predicted the performance of the membrane (MAPE of 10.5%) for the industrial RO brine. Overall, the models were able to successfully predict the water flux and can be used as potential process design tools.

Can consumer grade activity devices replace research grade actiwatches in youth mental health settings?

Mental health services increasingly recognize the benefits of monitoring sleep and activity patterns in youth. Objective, ecological monitoring via research grade actiwatches is valid, but an emerging literature suggests using cheaper, consumer grade activity tracking devices. This study aimed to compare the cross-validity of these devices, and their feasibility and usability in youth mental health settings. A convenience sample of 13 individuals recruited from youth mental health settings wore a consumer grade and research grade activity device for seven consecutive days. The five most frequently reported sleep and activity variables extracted directly from each device were compared using paired t tests, mean average percentage errors, and magnitude and direction of biases. Ten individuals provided usable recordings of sleep and activity parameters. Compared to the actiwatch recordings (criterion values), the consumer grade device showed systematic biases in the recording of total sleep time (overestimated by about an hour) and waking after sleep onset (underestimated by about half an hour). The consumer grade device was most reliable for monitoring daytime sedentary behaviour. However, participant feedback indicated that youth preferred the consumer grade device. The findings suggest that it is inappropriate to substitute actiwatches with commercial activity trackers in research setting, as the latter are probably insufficient for precision diagnostics or stratification of cases into treatment relevant subgroups. Consumer grade devices are more acceptable to youth, so they may be clinically useful for monitoring intra-individual where absolute accuracy is not required.

Short-term mixed electricity demand and price forecasting using adaptive autoregressive moving average and functional link neural network

A new hybrid adaptive autoregressive moving average (ARMA) and functional link neural network (FLNN) trained by adaptive cubature Kalman filter (ACKF) is presented in this paper for forecasting day-ahead mixed short-term demand and electricity prices in smart grids. The hybrid forecasting framework is intended to capture the dynamic interaction between the electricity consumers and the forecasted prices resulting in the shift of demand curve in electricity market. The proposed model comprises a linear ARMA-FLNN obtained by using a nonlinear expansion of the weighted inputs. The nonlinear functional block helps introduce nonlinearity by expanding the input space to higher dimensional space through basis functions. To train the ARMA-FLNN, an ACKF is used to obtain faster convergence and higher forecasting accuracy. The proposed method is tested on several electricity markets, and the performance metrics such as the mean average percentage error (MAPE), and error variance are compared with other forecasting methods, indicating the improved accuracy of the approach and its suitability for a real-time forecasting.

Enhancing privacy of electricity consumption in smart cities through morphing of anticipated demand pattern utilizing self-elasticity and genetic algorithms

The vision of smart cities has emerged from the integration of advances in information technologies to a city’s infrastructures and assets. The use of information will improve the daily operations, and lead to greener, safer, and more human friendly cities. The most preeminent city asset is the electrical power grid, upon which the modernization of human life is built. However, integration of power grid with information technologies, known as smart grid, comes at a cost of reduced privacy. Electricity consumer behavior expressed via daily consumption patterns may be visible to third parties, which can make critical inferences over the consumers’ personal life. In this paper, a new method is proposed that aims at enhancing consumer privacy in smart cities by proposing an intelligent aggregation of anticipated demand patterns of multiple consumers as a mean to hide individual features. To that end, the method utilizes consumers self-elasticities matrices and a genetic algorithm to create an aggregated pattern that masks individual consumption data. The proposed method is tested on real world electricity demand patterns, and morphing performance is recorded with respect to symmetric mean average percentage error (SMAPE) attaining morphing over 60% in all tested cases.

Forecasting of Potato Prices of Hooghly in West Bengal: Time Series Analysis Using SARIMA Model

Potato is one of the most important food crops in India. Potato is also one of the principal cash crop, it gives handsome returns to the farmers. West Bengal is a state in India where potato is one of the major agricultural crops. But potato markets are more volatile due to fluctuation in production rather than Stable. So price of potato is fluctuating in nature. Thus, the time series analysis and price forecast may help producers in acreage allocation and timing of sale of potato. The present study was conducted to know the statistical investigation of price behaviour of potato in Hooghly district of West Bengal. In the present study, Box-Jenkins Seasonal Auto Regressive Integrated Moving Average (SARIMA) modeling is deployed in forecasting of monthly average price of potato in Hooghly of West Bengal up to October 2020 based on data from November 2008 to October 2018 (a period of 120 months). Seasonal indices calculated showed that generally the price is low from January to April and it starts picking up from May and reaches the maximum in November. The best model has been selected based on the Bayesian Information Criteria (BIC), Root Means Square Error (RMSE), Mean Absolute Percent Error (MAPE), Mean Absolute Error (MAE) and highest R-Square. The estimated best SARIMA model is (1,1,0)(4,1,0)12. Root Mean Square Error (RMSE), Mean Average Percentage Error (MAPE), Mean Absolute Error (MAE) and BIC for Hooghly district are 184.10, 19.08, 118.98 and 10.69 respectively. Short term forecasts based on this model are close to the observed values and the behaviour of forecasted price of potato truly reflected the actual price as well as market tendency.

PERBANDINGAN INTERPOLASI DAN EKSTRAPOLASI NEWTON UNTUK PREDIKSIDATA TIME SERIES

For numerous purposes, time series data are analyzed to understand phenomena or behaviors of variables, and try to find future value. Interpolation is guessing time series data point between the range of data set. Extrapolation is predict or guessing time series data point from beyond the range of data set. &#x0D; In this study, Newton’s Extrapolation is compared with linear and squared extrapolation. Newton’s Extrapolation making the assumption that the observed trend continues for values of x outside the model range.&#x0D; The robustness of prediction using Root Mean Square Error (RMSE) and Mean Average Percentage Error (MAPE). The results of newton’s interpolation with bottom, middle, and top approaches found the best value are middle approach, namely RMSE 76,01 and MAPE 4,65%. In Newton’s Extrapolation, the error values are consistent at bottom, middle, and top approaches, namely RMSE 541,170 anda MAPE 33,19%. Based on data from the Statistics of Indonesia on the percentage and number of poor people in East Nusa Tenggara Province in 2010 -2018 is declining trend pattern. The error value with Linear, Quadratic, and Newton’s Extrapolation shows the robust value results at linear or trend extrapolation, namely RMSE 157,450 and MAPE 7,93%. These results indicate Newton's extrapolation works well on non-linear data and requires a combination method with soft computing methods such as Fuzzy Systems, AG, or ANN

Artificial intelligence techniques for the vibration, noise, and emission characteristics of a hydrogen-enriched diesel engine

ABSTRACTThe present paper investigates the prediction of vibration, noise level, and emission characteristics of a four-stroke, four-cylinder diesel engine fueled with sunflower, canola, and corn biodiesel blends while H2 injected through inlet manifold using two different artificial intelligence methods: artificial neural network (ANN) and support vector machines (SVM). The aim of using these methods is to predict vibration, noise, carbon monoxide (CO), CO2, and NOx based on the initial experimental study by varying engine speed, blends of biodiesel, and H2 energy substitution ratio. Experimental data were gathered from the literature. For the ANN method, LevenbergMarquardt backpropagation training algorithm with logarithmic sigmoid and linear transfer function for hidden and output layers, respectively, gives the best results for prediction of vibration, noise, and emission characteristics. For SVM, a regression model is implemented with Gaussian kernel function. Results show that the ANN performs better than SVM, and the best mean average percent error and R2 for the models developed are 2.03 and 0.988 for vibration acceleration, 0.39 and 0.9615 for noise, 7.27 and 0.8549 for CO, 5.09 and 0.9398 for NOx, and 2.21 and 0.993 for CO2 values, respectively. Eventually, it is found that the ANN method is a good choice for simulation and prediction of dual fueled hydrogen sunflower, canola, and corn biodiesel blends.

Virtual water quality monitoring at inactive monitoring sites using Monte Carlo optimized artificial neural networks: A case study of Danube River (Serbia)

Rationalization of water quality monitoring stations nowadays is applied in many countries. In some cases, missing data from abandoned/inactive stations, spatial and temporal, could be very important, hence the use of artificial neural networks (ANNs) for virtual water quality monitoring at inactive monitoring sites was investigated. The aim was to develop single-output and simultaneous ANNs for the spatial interpolation of 18 water quality parameters at single- and multi-inactive monitoring sites on Danube River course through Serbia. Those different modeling approaches were considered in order to determine the most suitable combination of models. The variable selection and sensitivity analysis in the case of simultaneous models were performed using a modified procedure based on Monte Carlo Simulations (MCS). In general, the multi-target models tend to be more accurate than single target ones, while single output models outperform the simultaneous ones. Hence, for particular monitoring network and set of water quality parameters the optimal combination of models must be defined based on model's accuracy and computational effort needed. The MCS selection procedure has proved to be efficient only in the case of simultaneous multi-target model. MCS based analysis of input-output interactions has shown all significant interactions in the case of simultaneous single-target are grouped as a complex cluster of interactions, where majority of inputs influence on several outputs. In the case multi-target model those interactions were portioned in five separate clusters, there majority of them mimic the input-output interactions that are present in single output models. The modeling strategy for study area was proposed on the basis of the performance of created models (mean average percentage error < 10%): simultaneous multi-target model for pH, alkalinity, conductivity, hardness, dissolved oxygen, HCO3−, SO42− and Ca, single-output multi-target models for temperature and Cl−, simultaneous single-target models for Mg and CO2, single output single target models for NO3−.

State estimation of resistive domestic hot water heaters in arbitrary operation modes for demand side management

Utilizing domestic hot water heaters (DHWH) for demand side management (DSM) requires physical models of DHWH dynamic behavior. An ideal dynamic DHWH model is simple as well as accurate. However, system identification and state estimation, which are necessary for real-life implementation, are usually not considered. This work reviews standard thermal models, such as the single-node model, the multi-node model, a two-mass composite model and a two-node hybrid model with thermocline tracking. A prediction error method (PEM), applicable to all four models, is developed. 16 weeks of user and switching data from a DSM field test are used to operate a laboratory DHWH realistically. The temperature distribution within the heater, as well as electrical power input, water volume flow, inlet and outlet water temperature, temperature of the surroundings, and thermal well temperature are recorded. System identification and state estimation routines are developed based on the data available for a retrofitted system. Using cross-validation on the available user and operation scenarios from the lab, the approaches are compared with respect to average and outlet temperature estimation errors, model robustness and computational costs. The results show that a single-node model, which incorporates information on stratification during heating intervals in assuming a linear temperature distribution in the lower part of the tank, performs nearly as well in predicting the average temperature (mean average percentage error of 6%) as the 50 times more computationally expensive multi-layer model (mean average percentage error of 4%) and better than all other single-layer model based approaches considered.

Group Method of Data Handling with Artificial Bee Colony in Combining Forecasts

In this study, the use of Artificial Intelligence (AI) algorithm to combine several time series forecasts is presented. This study is done by combining individual forecasts of Group Method of Data Handling models using the weighted-based combine approach. The weights for each individual model are calculated using Artificial Bee Colony algorithm. In order to evaluate the proposed model, this study tested the proposed model on the International Airline Passengers data, and the performances are calculated using mean square error (MSE), mean average error (MAE) and mean average percentage error (MAPE). The accuracy of the proposed model is compared to the individual models and the models implemented in previous literatures. The results revealed that the proposed model is able to produce significantly accurate forecasts

Cluster analysis and artificial neural networks in predicting energy efficiency of public buildings as a cost-saving approach

Abstract Although energy efficiency is a hot topic in the context of global climate change, in the European Union directives and in national energy policies, methodology for estimating energy efficiency still relies on standard techniques defined by experts in the field. Recent research shows a potential of machine learning methods that can produce models to assess energy efficiency based on available previous data. In this paper, we analyse a real dataset of public buildings in Croatia, extract their most important features based on the correlation analysis and chi-square tests, cluster the buildings based on three selected features, and create a prediction model of energy efficiency for each cluster of buildings using the artificial neural network (ANN) methodology. The main objective of this research was to investigate whether a clustering procedure improves the accuracy of a neural network prediction model or not. For that purpose, the symmetric mean average percentage error (SMAPE) was used to compare the accuracy of the initial prediction model obtained on the whole dataset and the separate models obtained on each cluster. The results show that the clustering procedure has not increased the prediction accuracy of the models. Those preliminary findings can be used to set goals for future research, which can be focused on estimating clusters using more features, conducted more extensive variable reduction, and testing more machine learning algorithms to obtain more accurate models which will enable reducing costs in the public sector.

Improvements in laser sheet dropsizing using numerical and experimental techniques

Laser sheet drop sizing (LSD) technique is used along with numerical tools to improve the accuracy of planar drop sizing in dense sprays. The multiple scattering in Mie and Planar Laser Induced Fluorescence (PLIF) signal is reduced using the Structured Laser Illumination Planar Imaging (SLIPI). The numerical corrections are applied to reduce error due to laser sheet scattering, absorption and auto-absorption of the PLIF signal. A novel methodology is proposed to measure the planar distribution of optical depth (OD) in airblast sprays using the LSD technique. The optical depth is observed to increase with an increase in Gas to Liquid mass Ratio (GLR) in air-blast sprays. It is demonstrated that, dependency of the LSD technique on another independent drop sizing technique can be minimized in airblast sprays using an empirical correlation. It is observed that the use of SLIPI improves the reliability of the LSD technique even in optically dilute sprays. The improvement in LSD techniques is significantly higher in dense sprays with the optical depth higher than 2.5. The maximum Mean Average Percentage Error (MAPE) reduced from 40% in conventional LSD technique to ∼12% for SLIPI-LSD with numerical corrections in dense sprays with OD ∼ 6. For the sprays with higher droplet density, it is necessary to use SLIPI in a combination with numerical correction techniques to improve the accuracy of the planar droplet sizing technique.

Performance Evaluation of Medium-Term Load Forecasting Approaches: A Case Study of Ogun State, Nigeria

The place of electrical energy in enhancement of this computer age cannot be over-emphasised. Its forecast plays a significant functions in energy industry, helps the government and private sectors in making the precise decision regarding energy management practices. This paper presents performance evaluation of medium-term load forecasting techniques: a case study of Ogun State, Nigeria. Two different approaches were used using the previous load consumption in 2017 for the forecast. Least square approach compared with regression exponential approaches gave the least value of Mean Average Percentage Error (MAPE) and Root Mean Square Error (RMSE), which are 1.8212% and 0.004472 respectively. The anticipated percentage load growth for the months of July-December, 2018 forecasted with least square approach were 34.06%, 33.54%, 36.10%, 31.10%, 32.23% and 30.15% respectively, acute gas supply caused by pipeline vandalisation and theft of distribution/sub-station materials could be held responsible for low load growth in the month of December. The results of this analysis will assist the Regional Headquarters, Ibadan Electricity Distribution Company (IBEDC), Abeokuta, Ogun State in making effective planning, operation and management of energy across the state.

Validation of an AutoRegressive Integrated Moving Average model for the prediction of animal zone temperature in a weaned piglet building

An AutoRegressive Integrated Moving Average model was validated for the prediction of temperatures in the animal zone of conventional weaned piglet barn. The validation period covered seven cycles and recorded values at 10-min intervals for 292 days. Average weight was 5.75 ± 0.86 kg at the beginning of the production cycle and 18.41 ± 2.12 kg at the end of the cycle. Mean outdoor air temperatures ranged 6.14 to 17.85 °C with deviations in the range 2.49 °C to 5.24 °C, which involved marked differences in the operation of the ventilation system. The Mean Average Percentage Error was below 4%, with a mean error of ≤1 °C. The Root Mean Square Error was in the range 0.77 °C to 1.19 °C, whereas the coefficient of determination ranged between 0.52 and 0.81. Despite the changes in environmental conditions and in animal weight and management, the accuracy of the model remained stable with low dispersion of values. The model showed good accuracy and reliability covering all the seasons under changing meteorological conditions because it considered the operation of the heating and ventilation systems and changes in animal weight. The residuals obtained from the validation of the seven production cycles were Gaussian distributed, which confirmed the validity of the model. The generated model can be used for more effective environmental control systems that are capable of anticipating events and show a better response, which helps improve energy savings and animal welfare.

Prediksi Hasil Ujian Kompetensi Mahasiswa Program Profesi Dokter (UKMPPD) dengan Pendekatan ANFIS

Tujuan utama dari penelitian ini adalah untuk memprediksi hasil ujian kompetensi mahasiswa program profesi dokter (UKMPPD) menggunakan metode Adaptive Neuro Fuzzy Inference System (ANFIS). Data didapatkan dari database mahasiswa Fakultas Kedokteran Universitas Riau tahun 2015 yang berjumlah 170 data. Variabel input yang digunakan meliputi status kepesertaan, lama studi, dan Indeks Prestasi Kumulatif. Selanjutnya data dianalisis menggunakan software MATLAB dengan pengaturan jumlah membership function 2 2 2 dan type membership funtion gbell. Hasil penelitian menunjukkan bahwa metode ANFIS mampu memprediksi hasil UKMPPD dengan nilai Mean Average Percentage Error (MAPE) sebesar 0,07%, minimal 0,00% dan maximal 0,44%.

Occupancy prediction through machine learning and data fusion of environmental sensing and Wi-Fi sensing in buildings

Abstract Occupancy information is crucial to building facility design, operation, and energy efficiency. Many studies propose the use of environmental sensors (such as carbon dioxide, air temperature, and relative humidity sensors) and radio-frequency sensors (Wi-Fi networks) to monitor, assess, and predict occupancy information for buildings. As many methods have been developed and a variety of sensory data sources are available, establishing a proper selection of model and data source is critical to the successful implementation of occupancy prediction systems. This study compared three popular machine learning algorithms, including k-nearest neighbors (kNN), support vector machine (SVM), and artificial neural network (ANN), combined with three data sources, including environmental data, Wi-Fi data, and fused data, to optimize the occupancy models' performance in various scenarios. Three error measurement metrics, the mean average error (MAE), mean average percentage error (MAPE), and root mean squared error (RMSE), have been employed to compare the models' accuracies. Examined with an on-site experiment, the results suggest that the ANN-based model with fused data has the best performance, while the SVM model is more suitable with Wi-Fi data. The results also indicate that, comparing with independent data sources, the fused data set does not necessarily improve model accuracy but shows a better robustness for occupancy prediction.

Very Sort Term Load Forecasting Using Interval Type - 2 Fuzzy Inference System (IT- 2 FIS) (Case Study: Java Bali Electrical System)

One of the important things to do an electric power system operation is load forecasting. Load forecasting consists of short-term forecasting and short-term forecasting. The very short term load forecasting are required for regulating electrical energy generation, maintenance arrangements and regulating the labor involved. This forecasting is done to decide which plant to operate. The capacity of the plant to be operated adjusts to the load plan to be supplied the next day. The very short-term load forecasting is predicting electrical loads with time intervals every 30 minutes for the next day. In this study using Interval Type-2 Fuzzy Inference System (IT-2FIS) because it delivers a high flexibility that can be developed using other methods (hybrid). Laying out the footprint of uncertainty (FOU) membership function of the Interval Type-2 Fuzzy Inference System (IT-2FIS). This method has been applied for short- term load forecasting and will be employed for very short-term forecasting. In very short-term load forecasting IT-2 FIS has Mean Average Percentage Error (MAPE) arround 0,729%.