Autoregressive Neural Network Research Articles

Financial time series prediction using wavelet and artificial neural network

This paper focuses on the modelling of financial time series using the coupling of the discrete wavelet transform and nonlinear autoregressive neural network. This hybrid modelling method is based on the use of decomposed time series using the discrete wavelet transform as inputs to the artificial neural networks. This method has been applied to three financial series (exchange rate EUR/USD, the Brent price and NASDAQ composite price). The simulation results using R software show the robustness of the proposed model compared to other modelling methods applied to the same financial time series.

Predictive analysis of the number of human brucellosis cases in Xinjiang, China

Brucellosis is one of the major public health problems in China, and human brucellosis represents a serious public health concern in Xinjiang and requires a prediction analysis to help making early planning and putting forward science preventive and control countermeasures. According to the characteristics of the time series of monthly reported cases of human brucellosis in Xinjiang from January 2008 to June 2020, we used seasonal autoregressive integrated moving average (SARIMA) method and nonlinear autoregressive regression neural network (NARNN) method, which are widely prevalent and have high prediction accuracy, to construct prediction models and make prediction analysis. Finally, we established the SARIMA((1,4,5,7),0,0)(0,1,2)12 model and the NARNN model with a time lag of 5 and a hidden layer neuron of 10. Both models have high fitting performance. After comparing the accuracies of two established models, we found that the SARIMA((1,4,5,7),0,0)(0,1,2)12 model was better than the NARNN model. We used the SARIMA((1,4,5,7),0,0)(0,1,2)12 model to predict the number of monthly reported cases of human brucellosis in Xinjiang from July 2020 to December 2021, and the results showed that the fluctuation of the time series from July 2020 to December 2021 was similar to that of the last year and a half while maintaining the current prevention and control ability. The methodology applied here and its prediction values of this study could be useful to give a scientific reference for prevention and control human brucellosis.

Soft Computing of a Medically Important Arthropod Vector with Autoregressive Recurrent and Focused Time Delay Artificial Neural Networks.

Simple SummaryArthropod vectors are responsible for transmitting a large number of diseases, and for most, there are still not available effective vaccines. Vector disease control is mostly achieved by a sustained prediction of vector populations to maintain support for surveillance and control activities. Mathematical models may assist in predicting arthropod population dynamics. However, arthropod dynamics, and mosquitoes particularly, due their complex life cycle, often exhibit an abrupt and non-linear occurrence. Therefore, there is a growing interest in describing mosquito population dynamics using new methodologies. In this work, we made an effort to gain insights into the non-linear population dynamics of Culex sp. adults, aiming to introduce straightforward soft-computing techniques based on artificial neural networks (ANNs). We propose two kind of models, one autoregressive, handling temperature as an exogenous driver and population as an endogenous one, and a second based only on the exogenous factor. To the best of our knowledge, this is the first study using recurrent neural networks and the most influential environmental variable for prediction of the WNv vector Culex sp. population dynamics, providing a new framework to be used in arthropod decision-support systems.A central issue of public health strategies is the availability of decision tools to be used in the preventive management of the transmission cycle of vector-borne diseases. In this work, we present, for the first time, a soft system computing modeling approach using two dynamic artificial neural network (ANNs) models to describe and predict the non-linear incidence and time evolution of a medically important mosquito species, Culex sp., in Northern Greece. The first model is an exogenous non-linear autoregressive recurrent neural network (NARX), which is designed to take as inputs the temperature as an exogenous variable and mosquito abundance as endogenous variable. The second model is a focused time-delay neural network (FTD), which takes into account only the temperature variable as input to provide forecasts of the mosquito abundance as the target variable. Both models behaved well considering the non-linear nature of the adult mosquito abundance data. Although, the NARX model predicted slightly better (R = 0.623) compared to the FTD model (R = 0.534), the advantage of the FTD over the NARX neural network model is that it can be applied in the case where past values of the population system, here mosquito abundance, are not available for their forecasting.

Hybrid Model for Time Series of Complex Structure with ARIMA Components

A hybrid model for the time series of complex structure (HMTS) was proposed. It is based on the combination of function expansions in a wavelet series with ARIMA models. HMTS has regular and anomalous components. The time series components, obtained after expansion, have a simpler structure that makes it possible to identify the ARIMA model if the components are stationary. This allows us to obtain a more accurate ARIMA model for a time series of complicated structure and to extend the area for application. To identify the HMTS anomalous component, threshold functions are applied. This paper describes a technique to identify HMTS and proposes operations to detect anomalies. With the example of an ionospheric parameter time series, we show the HMTS efficiency, describe the results and their application in detecting ionospheric anomalies. The HMTS was compared with the nonlinear autoregression neural network NARX, which confirmed HMTS efficiency.

Comparison of algorithms for the prediction of glucose levels in patients with diabetes

This work presents a comparison between two algorithms for the prediction of glucose levels in diabetic patients by using a univariate time series. The algorithms are applied to the history of fasting glucose levels to predict the five following values. The comparison is performed between 1) The Autoregressive Neural Networks (ARNN) and 2) The autoregressive integrated moving average (ARIMA) models. A total of 70 series are analyzed, and we show that the results obtained for the ARIMA model have error percentages higher than 25% of the predicted value to the expected value. In contrast, in 73% of the cases, the percentage error was less than 25% for the Autoregressive Neural Networks.

Modeling current and future role of agricultural waste in the production of bioethanol for gasoline vehicles

This study addressed the urgent need for biofuels even in countries such as Iran, which has fairly good fossil fuel resources. The problems, necessities of the present, and future demand for biofuels are discussed. As the transportation sector is one of the largest sources of air pollution, this study has focused on this sector. This issue was examined from a global perspective, and then within the context of domestic bioethanol production using agricultural residues and proposing different scenarios. The first step in implementing this policy is the accurate forecast of the demand for second-generation bioethanol in the coming years. A nonlinear auto regressive neural network was applied to predict gasoline demand based on Mackey-Glass chaotic time series. Gasoline demand is forecasted by 2030, based on projected volumes of gasoline in different bioethanol mixture scenarios. Results revealed that using scenarios E10, E15, E25, the volume of bioethanol needed by 2030 will amount to 10.12, 15.16 and 25.31 million L per day which can be produced using agricultural products wastes.

Double ensemble system for wind energy forecasting based on generalized autoregressive conditional heteroskedasticity and neural network models with variational mode decomposition

ABSTRACT With the steady integration of wind energy into electricity networks, precise wind speed forecasting is an essential element in the administration and management of power systems. However, wind energy forecasting research has focused increasingly on short-term forecasting, leaving aside the challenging horizons of medium- and long-term predictions. Therefore, this study proposes a wind speed forecasting methodology based on two types of ensembles, which addresses the nonlinearity and chaotic behavior of wind speed using decomposition-based models. With the results of the first ensemble of 90 ARMA-generalized autoregressive conditional heteroskedasticity (ARMA-GARCH) models, the second ensemble is established based on three types of neural networks and learning functions. Finally, we propose the application of variational mode decomposition (VMD) before or after the first ensemble. The experimental outcomes lead us to divide the prediction horizons into two broad groups, those where VMD inclusion did and did not improve the ensemble results. These horizons are classified as short-term (3, 4, and 5 steps) and mid- and long-term forecast horizons (6, 12, 24, and 48 steps), where the best performance arises with the VMD application after the first ensemble. The research contributes to the existing literature studying a wide variety of innovation distribution and optimization methods that can be implemented with GARCH-type models. Simultaneously, the VMD application is proposed in a novel way not seen in the literature by applying it to the predictions already made by other models, in this case, in ensembles of GARCH-type models.

Predicting temperature of Erbil City applying deep learning and neural network

<span>One of the most significant and daunting activities in today's world is temperature prediction. The meteorologists traditionally predict temperature via some statistical models aimed to forecast the fluctuations that might have happened to atmospheric parameters such as temperature, humidity, etc. The main objective of this paper is to build an intelligent temperature prediction model of Erbil city in KRG/ Iraq based on a historical dataset from 1992 to 2016 in each year there are twelve months’ average temperature readings from (January to December). Hence to resolve this prediction problem an up-to-date deep learning neural network has been used, the network model is based on <span id="docs-internal-guid-850bd062-7fff-c6c8-9146-ba6427eb24e0" style="font-size: 9pt; font-family: 'Times New Roman'; color: #000000; background-color: transparent; font-weight: 400; font-style: normal; font-variant: normal; text-decoration: none; vertical-align: baseline; white-space: pre-wrap;">long short-term memory</span> (LSTM) as an artificial recurrent neural network (RNN) architecture which employed to estimate the future average temperature. The implementing model uses the dataset from real-time 30 weather stations deployed in the area of the city. The prediction performance of the proposed recurrent neural network model has been compared with some state of art algorithms like Adeline neural network, Autoregressive neural network (NAR), and <span id="docs-internal-guid-14d37b98-7fff-0f76-848f-ad9f89224f77" style="font-size: 9pt; font-family: 'Times New Roman'; color: #000000; background-color: transparent; font-weight: 400; font-style: normal; font-variant: normal; text-decoration: none; vertical-align: baseline; white-space: pre-wrap;"> generalized regression neural network</span> (GRNN). The results show that the proposed model based on deep learning gives minimum prediction error.</span>

Forecasting the Tuberculosis Incidence Using a Novel Ensemble Empirical Mode Decomposition-Based Data-Driven Hybrid Model in Tibet, China.

ObjectiveThe purpose of this study is to develop a novel data-driven hybrid model by fusing ensemble empirical mode decomposition (EEMD), seasonal autoregressive integrated moving average (SARIMA), with nonlinear autoregressive artificial neural network (NARNN), called EEMD-ARIMA-NARNN model, to assess and forecast the epidemic patterns of TB in Tibet.MethodsThe TB incidence from January 2006 to December 2017 was obtained, and then the time series was partitioned into training subsamples (from January 2006 to December 2016) and testing subsamples (from January to December 2017). Among them, the training set was used to develop the EEMD-SARIMA-NARNN combined model, whereas the testing set was used to validate the forecasting performance of the model. Whilst the forecasting accuracy level of this novel method was compared with the basic SARIMA model, basic NARNN model, error-trend-seasonal (ETS) model, and traditional SARIMA-NARNN mixture model.ResultsBy comparing the accuracy level of the forecasting measurements including root-mean-square error, mean absolute deviation, mean error rate, mean absolute percentage error, and root-mean-square percentage error, it was shown that the EEMD-SARIMA-NARNN combined method produced lower error rates than the others. The descriptive statistics suggested that TB was a seasonal disease, peaking in late winter and early spring and a trough in autumn and early winter, and the TB epidemic indicated a drastic increase by a factor of 1.7 from 2006 to 2017 in Tibet, with average annual percentage change of 5.8 (95% confidence intervals: 3.5–8.1).ConclusionThis novel data-driven hybrid method can better consider both linear and nonlinear components in the TB incidence than the others used in this study, which is of great help to estimate and forecast the future epidemic trends of TB in Tibet. Besides, under present trends, strict precautionary measures are required to reduce the spread of TB in Tibet.

Modeling and Forecasting Long-Term Records of Mean Sea Level at Grand Isle, Louisiana: SARIMA, NARNN, and Mixed SARIMA-NARNN Models

This study tried to demonstrate the role of time series models in modeling and forecasting process using long-term records of monthly mean sea level from January 1978 to October 2020 at Grand Isle, Louisiana. Following the Box–Jenkins methodology, the ARIMA(1,1,1)(2,0,0)12 with drift model was selected to be the best fit model for the time series, according to its lowest AIC value. Using the LM algorithm, the results revealed that the NARNN model with 9 neurons in the hidden layer and 6 time delays provided the best performance in the nonlinear autoregressive neural network models at its smaller MSE value. The Mixed model, a combination of the SARIMA and NARNN models has both linear and nonlinear modelling capabilities can be a better choice for modelling the time series. The comparative results revealed that the Mixed-LM model with 9 neurons in the hidden layer and 3 time delays yielded higher accuracy than the NARNN-LM model with 9 neurons in the hidden layer and 6 time delays, and the ARIMA(1,1,1)(2,0,0)12 with drift model, according to its lowest MSE in this study. Thus, this study may provide an integrated modelling approach as a decision-making supportive method for formulating local mean sea level forecast in advance. Understanding past sea level is important for the analysis of current and future sea level changes. In order to sustain these observations, research programs utilizing the resulting data should be able to improve significantly our understanding and narrow projections of future sea level rise and variability.

Integrate new cross association fuzzy logical relationships to multi-factor high-order forecasting model of time series

In any multi-factor high-order fuzzy logical relationship (FLR) based forecasting model, a FLR reflects the influence of both the main factor (the forecasted factor) and all the influence factors on the main factor. Thus, the antecedent of a FLR includes multiple premises related to the main factor as well as all the influence factors. In real time series, there may exist another kind of influence: the cross association influence which is from a part of influence factor(s) on the main factor. To describe such kind of influence, we propose the concept of multi-factor high-order cross association FLRs (CAFLRs). The antecedent of a CAFLR includes some premises related to a part of influence factors. The proposed CAFLRs are divided into two categories: short-cross association FLRs and long-cross association FLRs, which describe the influence on the consequent observation from the premise observations at the closest consecutive moments and the premise observations at the non-closest non-consecutive moments respectively. Based on the concept of CAFLRs, a novel forecasting model is built up. In the proposed model, more FLRs than in the existing models can be mined from historical observations and added to the rule base, which further improve the prediction accuracy by raising the possibility of finding available forecasting FLRs. Superior performance of the proposed model has been verified in the experiments by comparing with Nonlinear Autoregressive Neural Networks, Autoregressive Model, Support Vector Regression and some other FLR based forecasting models.

Motion artifact synthesis for research in biomedical signal quality analysis

Advances in microprocessors, sensor technology, and signal processing have led to the increased feasibility of biosignal-integrated wearable smart devices for long-term monitoring, mobility assistance, and fitness and sports analytics. Motion artifact contamination can adversely affect the interpretation of bioelectric signals (e.g., electrocardiograms, electromyograms) collected by these wearable devices. Research and development of biomedical signal quality analysis algorithms to detect, identify, quantify, and mitigate motion artifact is typically performed using a ground truth signal contaminated with previously recorded or simulated motion artifact. The procedures for generating simulated motion artifact are undercharacterized in the literature, and many research simulations have been based on the same recording of motion artifact data, leading to concerns about the diversity of the simulated data. This paper proposes and compares three methods for synthesizing motion artifact data. Motion artifact data was collected from 5 subjects performing two different tasks (arm movement and walking). Autoregressive (AR), Markov chain (MC), and recurrent neural network (RNN) models were explored to generate simulated motion artifact data. The AR model imitated time and frequency domain properties effectively, but was ineffective for reproducing the morphology and probability distribution. The MC model was able to produce data with similar time domain properties and probability distribution, but was less effective than the RNN model at imitating the frequency domain and morphology. We conclude that the RNN method is promising for synthesizing diverse motion artifact data that resembles the properties of experimental data to achieve a ground truth signal for research in bioelectric signal quality analysis.

Overview, modeling and forecasting the effects of COVID-19 pandemic on energy market and electricity demand: a case study on Turkey

ABSTRACT In this study, the overall energy status of Turkey in the early stage of the COVID-19 pandemic was analyzed. The situation in the energy market and electricity demand in the pandemic period as global issues were investigated in Turkey as a developing country. In addition to this general overview using limited data for the early period of the pandemic, energy demand and energy generation values were modeled utilizing machine learning approaches. Daily energy demand values were modeled and forecasted by utilizing Nonlinear Autoregression Neural Network (NARNN), Auto-Regressive Integrated Moving Average (ARIMA), and Long-Short Term Memory (LSTM) techniques. According to the results of the first stage of the modeling process, the LSTM approach was found as the most accurate model. In the second step of the modeling and forecasting analysis, monthly electricity generation values from natural gas and coal were predicted. In the energy generation forecasting analysis for April-December 2020 period, contraction in energy generation by natural gas and coal were obtained as 6.27% and 7.19%, respectively in comparison to 2019. Finally, the energy market during and after the pandemic was evaluated and the strategies to be implemented in the post-pandemic process were discussed.

Improving Quality of Long-Term Bond Price Prediction Using Artificial Neural Networks

Purpose: The aim of this paper is to propose nonlinear autoregressive neural network which can improve quality of bond price forecasting. Methodology/Approach: Due to the complex nature of market information that influence bonds, artificial intelligence could be accurate, robust and fast choice of bond price prediction method. Findings: Our results have reached a coefficient of determination higher than 95% in the training, validation and testing sets. Moreover, we proposed the nonlinear autoregressive network with external inputs using 50 year interest-rate swaps denominated in EUR and volatility index VIX as two external variables. Research Limitation/Implication: Our sample of daily prices between 4th January 2016 and 13th January 2021 (totally 1,270 trading days) suggest that both Levenberg-Marquardt and Scaled conjugate gradient learning algorithms achieved excellent results. Originality/Value of paper: Despite the fact that both learning algorithms achieved satisfying outcomes, implementation of an independent variable into the autoregressive neural network environment had no significant impact on prediction ability of the model.

Nonlinear Neural Network Based Forecasting Model for Predicting COVID-19 Cases.

The recent COVID-19 outbreak has severely affected people around the world. There is a need of an efficient decision making tool to improve awareness about the spread of COVID-19 infections among the common public. An accurate and reliable neural network based tool for predicting confirmed, recovered and death cases of COVID-19 can be very helpful to the health consultants for taking appropriate actions to control the outbreak. This paper proposes a novel Nonlinear Autoregressive (NAR) Neural Network Time Series (NAR-NNTS) model for forecasting COVID-19 cases. This NAR-NNTS model is trained with Scaled Conjugate Gradient (SCG), Levenberg Marquardt (LM) and Bayesian Regularization (BR) training algorithms. The performance of the proposed model has been compared by using Root Mean Square Error (RMSE), Mean Square Error (MSE) and correlation co-efficient i.e. R-value. The results show that NAR-NNTS model trained with LM training algorithm performs better than other models for COVID-19 epidemiological data prediction.

A New Unit Root Test for Unemployment Hysteresis Based on the Autoregressive Neural Network*

AbstractThis paper proposes a nonlinear unit root test based on the autoregressive neural network process for testing unemployment hysteresis. In this new unit root testing framework, the linear, quadratic and cubic components of the neural network process are used to capture the nonlinearity in a given time series data. The theoretical properties of the test are developed, while the size and the power properties are examined in a Monte Carlo simulation study. Various empirical applications with unemployment and inflation rates across a number of countries are carried out at the end of the article.

Adoption of bio-image technology on rehabilitation intervention of sports injury of golf

This work aimed to explore the characteristics of surface electromyography (EMG) signal of golfers’ low back pain and the effect of rehabilitation. Based on the time-varying parameter autoregressive model and artificial neural network, ARAN algorithm was constructed, which was compared with the autoregressive moving average (ARMA) algorithm and the convolutional neural network (CNN) algorithm. Then, the established ARAN algorithm was employed to evaluate the characteristics of surface EMG signal of 106 golfers with low back pain. It was found that the accuracy, sensitivity, and specificity of the ARAN algorithm were superior to those of the CNN and ARMA algorithms. The golfer’s Roland-Morris Disability Questionnaire (RMDQ) score after treatment was less than that before treatment (P < 0.05). Moreover, there was significant negative correlation between RMDQ score and the mean values of time-varying parameters a1 and a3 (P < 0.05). The RMDQ score had a very obvious positive correlation with the mean values of a2, a4, and a6 (P < 0.001) and had a negative correlation with the mean value of a5 (P < 0.05). To sum up, the time-varying parameters of the surface EMG signal can effectively evaluate the golfer’s low back pain and the effect of treatment and rehabilitation.

Improving the precision of modeling the incidence of hemorrhagic fever with renal syndrome in mainland China with an ensemble machine learning approach.

Hemorrhagic fever with renal syndrome (HFRS), one of the main public health concerns in mainland China, is a group of clinically similar diseases caused by hantaviruses. Statistical approaches have always been leveraged to forecast the future incidence rates of certain infectious diseases to effectively control their prevalence and outbreak potential. Compared to the use of one base model, model stacking can often produce better forecasting results. In this study, we fitted the monthly reported cases of HFRS in mainland China with a model stacking approach and compared its forecasting performance with those of five base models. We fitted the monthly reported cases of HFRS ranging from January 2004 to June 2019 in mainland China with an autoregressive integrated moving average (ARIMA) model; the Holt-Winter (HW) method, seasonal decomposition of the time series by LOESS (STL); a neural network autoregressive (NNAR) model; and an exponential smoothing state space model with a Box-Cox transformation; ARMA errors; and trend and seasonal components (TBATS), and we combined the forecasting results with the inverse rank approach. The forecasting performance was estimated based on several accuracy criteria for model prediction, including the mean absolute percentage error (MAPE), root-mean-squared error (RMSE) and mean absolute error (MAE). There was a slight downward trend and obvious seasonal periodicity inherent in the time series data for HFRS in mainland China. The model stacking method was selected as the best approach with the best performance in terms of both fitting (RMSE 128.19, MAE 85.63, MAPE 8.18) and prediction (RMSE 151.86, MAE 118.28, MAPE 13.16). The results showed that model stacking by using the optimal mean forecasting weight of the five abovementioned models achieved the best performance in terms of predicting HFRS one year into the future. This study has corroborated the conclusion that model stacking is an easy way to enhance prediction accuracy when modeling HFRS.

Forecasting Betula and Poaceae airborne pollen concentrations on a 3-hourly resolution in Augsburg, Germany: toward automatically generated, real-time predictions

Airborne allergenic pollen impact the health of a great part of the global population. Under climate change conditions, the abundance of airborne pollen has been rising dramatically and so is the effect on sensitized individuals. The first line of allergy management is allergen avoidance, which, to date, is by rule achieved via forecasting of daily pollen concentrations. The aim of this study was to elaborate on 3-hourly predictive models, one of the very few to the best of our knowledge, attempting to forecast pollen concentration based on near-real-time automatic pollen measurements. The study was conducted in Augsburg, Germany, during four years (2016–2019) focusing on Betula and Poaceae pollen, the most abundant and allergenic in temperate climates. ARIMA and dynamic regression models were employed, as well as machine learning techniques, viz. artificial neural networks and neural network autoregression models. Air temperature, relative humidity, precipitation, air pressure, sunshine duration, diffuse radiation, and wind speed were additionally considered for the development of the models. It was found that air temperature and precipitation were the most significant variables for the prediction of airborne pollen concentrations. At such fine temporal resolution, our forecasting models performed well showing their ability to explain most of the variability of pollen concentrations for both taxa. However, predictive power of Betula forecasting model was higher achieving R2 up to 0.62, whereas Poaceae up to 0.55. Neural autoregression was superior in forecasting Betula pollen concentrations, whereas, for Poaceae, seasonal ARIMA performed best. The good performance of seasonal ARIMA in describing variability of pollen concentrations of both examined taxa suggests an important role of plants’ phenology in observed pollen abundance. The present study provides novel insight on per-hour forecasts to be used in real-time mobile apps by pollen allergic patients. Despite the huge need for real-time, short-term predictions for everyday clinical practice, extreme weather events, like in the year 2019 in our case, still comprise an obstacle toward highly performing forecasts at such fine timescales, highlighting that there is still a way to go to this direction.

Forecasting emergency department admissions

The emergency department of a hospital plays an extremely important role in the healthcare of patients. To maintain a high quality service, clinical professionals need information on how patient flow will evolve in the immediate future. With accurate emergency department forecasts it is possible to better manage available human resources by allocating clinical staff before peak periods, thus preventing service congestion, or releasing clinical staff at less busy times. This paper describes a solution developed for the presentation of hourly, four-hour, eight-hour and daily number of admissions to a hospital’s emergency department. A 10-year history (2009-2018) of the number of emergency admissions in a Portuguese hospital was used. To create the models several methods were tested, including exponential smoothing, SARIMA, autoregressive and recurrent neural network, XGBoost and ensemble learning. The models that generated the most accurate hourly time predictions were the recurrent neural network with one-layer (sMAPE = 23.26%) and with three layers (sMAPE = 23.12%) and XGBoost (sMAPE = 23.70%). In terms of efficiency, the XGBoost method has by far outperformed all others. The success of the recurrent neuronal network and XGBoost machine learning methods applied to the prediction of the number of emergency department admissions has been demonstrated here, with an accuracy that surpasses the models found in the literature.