Search

Accurate prediction is a fundamental and leading work of the emergency medicine reserve management. Given that the emergency medicine reserve demand is affected by various factors during the public health events and thus the observed data are composed of different but hard-to-distinguish components, the traditional demand forecasting method is not competent for this case. To bridge this gap, this paper proposes the EMD-ELMAN-ARIMA (ELA) model which first utilizes Empirical Mode Decomposition (EMD) to decompose the original series into various components. The Elman neural network and ARIMA models are employed to forecast the identified components and the final forecast values are generated by integrating the individual component predictions. For the purpose of validation, an empirical study is carried out based on the influenza data of Beijing from 2014 to 2018. The results clearly show the superiority of the proposed ELA algorithm over its two rivals including the ARIMA and ELMAN models.

Purpose Coronavirus disease is an irresistible infection caused by the respiratory disease coronavirus 2 (SARS-CoV-2). It was first found in Wuhan, China, in December 2019, and has since spread universally, causing a constant pandemic. On June 3, 2020, 6.37 million cases were found in 188 countries and regions. During pandemic prevention, this can minimize the impact of the disease on individuals and groups. A study was carried out on coronavirus to observe the number of cases, deaths, and recovery cases worldwide within a specific time period of 5 months. Based on this data, this research paper will predict the future spread of this infectious disease in human society. Methods In our study, the dataset was taken from WHO “Data WHO Coronavirus Covid-19 cases and deaths-WHO-COVID-19-global-data”. This dataset contains information about the observation date, provenance/state, country/region, and latest updates. In this article, we implemented several forecasting techniques: naive method, simple average, moving average, single exponential smoothing, Holt linear trend method, Holt-Winters method and ARIMA, for comparison, and how these methods improve the Root mean square error score. Results The naive method is best suited as described over all other methods. In the ARIMA model, utilizing grid search, we recognized a lot of boundaries that delivered the best-fit model for our time series data. By continuing the model, future predictions of death cases indicate that the number of deaths will increased by more than 600,000 by January 2021. Conclusion This survey will support the government and experts in making arrangements for what is about to happen. Based on the findings of instantaneous model, these models can be adjusted to guide long time.

For almost a century, classical statistical methods including exponential smoothing and autoregression integrated moving averages (ARIMA) have been predominant in the analysis of time series (TS) and in the pursuit of forecasting future events from historical data. TS are chronological sequences of observations, and TS data are therefore prevalent in many aspects of clinical medicine and academic neuroscience. With the rise of highly complex and nonlinear datasets, machine learning (ML) methods have become increasingly popular for prediction or pattern detection and within neurosciences, including neurosurgery. ML methods regularly outperform classical methods and have been successfully applied to, inter alia, predict physiological responses in intracranial pressure monitoring or to identify seizures in EEGs. Implementing nonparametric methods for TS analysis in clinical practice can benefit clinical decision making and sharpen our diagnostic armory.

Air pollution causes a number of pulmonary and cardiovascular diseases. Recording of air pollution via real-time low-cost IoT-based monitoring systems and its subsequent forecasting are likely to help timely warn people about prevailing air pollution across a large number of sites. In this paper, we propose and compare a real-time low-cost IoT-based air pollution monitoring system against an existing, accurate, and expensive industry-grade system. Furthermore, we undertake the task of predicting the accurate values of the industry-grade system from values recorded by the low-cost system. For forecasting, a Vector Autoregressive (VAR) model, a Vector Autoregressive Moving Average (VARMA) model, a Seasonal Autoregressive Integrated Moving Average with Exogenous variable (SARIMAX) model, and a weighted ensemble model of VAR, VARMA, and SARIMAX models were trained and tested on particular matter data. Data for forecasting were collected from the low-cost monitoring system and the industry-grade system over a period of time. Results revealed that the low-cost monitoring system predicted the values of the industry-grade system accurately. Furthermore, the ensemble model performed the best among all models in forecasting of accurate particular matter values of the industry-grade system by using the output of the low-cost system. We highlight the implication of using low-cost systems for monitoring of air pollution.

In this work, a novel bidirectional model based on the recurrent neural network known as Gated Recurrent Unit (GRU) is proposed for the imputation of not available (NA) values ​​in daily wind speed time series. The proposal model consists of two sequential GRU sub-models of 4-layers each, and for experimentation data from 3 years (2018–2020) is used, the first sub-model is trained with data from 2018 and the second sub-model with 2020 data, in both cases 2019 data is predicted, also, for second sub-model it's necessary the flipped 2020 data. Likewise, data augmentation is applied to improve the precision of the NA estimations. The results achieved show that the bidirectional proposal model achieves very good results, outperforming benchmark models such as Local Average of Nearest Neighbors (LANN), Autoregressive Integrated Moving Average (ARIMA) and Gated Recurrent Unit (GRU) without data augmentation. Likewise, comparing the results with other related works, it’s observed that proposal model surpasses most of them, making it an excellent alternative for wind speed time series imputation.

Fast spreading coronavirus disease 2019 (COVID-19), originated in the Wuhan city, China in December 2019, is a contagious disease caused by Severe Acute Respiratory Syndrome-Coronavirus-2 (SARS-CoV-2). Within a short period, COVID-19 infections spread over large geographical area affecting millions of people and declared a pandemic by the World Health Organization (WHO). The fast and quick spread of the virus across the globe resulted in thousands of casualties. COVID-19 prevalence in India was reported in the late of January 2020 and the number of infections increased sharply by the end of March. In such a troublesome situation, time series analysis proves very much helpful in monitoring and assessing the growth curve of COVID-19 infections. In the present study, autoregressive integrated moving average (ARIMA) models are developed for the time series data of cumulative confirmed, recovered and causalities cases of COVID-19 in India. The data set under study is broken up into two subsets, modelling and testing data sets. After analysing the input data for stationarity using autocorrelation function (ACF) and partial correlation function (PACF) plots, different ARIMA models are estimated for confirmed, recovered and causalities’ cases of COVID-19 in India for modelling phase. ARIMA Model outputs are then compared with observed values of confirmed, recovered and casualties’ cases for the testing phase using error analysis. It has been found that ARIMA \(\left( {0,2,3} \right)\), ARIMA \(\left( {0,2,5} \right)\) and ARIMA \(\left( {1,2,1} \right)\) models are appropriate with the lowest mean absolute percentage error (MAPE) values for the data of confirmed cases, recovered cases and casualties’ cases respectively. Finally, the developed ARIMA models are used to forecast one-month ahead values of confirmed, recovered and casualties’ cases of COVID-19 in India. The predictions indicate rise in confirmed COVID-19 cases and speedy recoveries as well, whereas the casualties continue to show a constant trend in future. Based on these future trends of COVID-19 outbreak, governments and policymakers can take preventive measures to break the ongoing chain of COVID-19 infections and make necessary arrangements in the wake of an emergency.

The world at large has been confronted with several disease outbreaks which have posed and still posing a serious menace to public health globally. Recently, COVID-19 a new kind of coronavirus emerge from Wuhan city in China and was declared a pandemic by the World Health Organization. There has been reported case of about 10,021,401 with global death of 499,913 as of 15.15 GMT, June 29 2020. There are 382,190 and 9664 positive cases and deaths in Africa, respectively as of June 29 at 7:00 GMT. South-Africa, Egypt, Nigeria, Ghana, Algeria and Cameroon are the most affected African countries with this outbreak. The chapter referred to them in this study as Africa epicenters’. Thus, there is a need to monitor and predict COVID-19 prevalence in this region for effective control and management. Different statistical tools and time series model such as the linear regression model and auto-regressive integrated moving average (ARIMA) models have been applied for disease prevalence/incidence prediction in different diseases outbreak. This book chapter adopted the linear regression model and the ARIMA models to forecast the trend of COVID-19 prevalence in the aforementioned African countries. The datasets examined in this analysis spanned from February 21, 2020, to October 4, 2020, and was extracted from the European Centre for Disease Prevention and Control website. The cumulative confirmed cases of COVID-19 cases were subjected to different curve estimation statistical models in simple, quadratic, cubic, and quartic forms. In the chapter, we identified the best model in each country and use the same for prediction and forecasting purposes. In conclusion, we obtained the future trend of this virus across Africa epicenters’ and this, in turn, will assist the government and health authorities to plan and take precautions that will help to curb this pandemic in Africa. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.

The spreading and Development of COVID-19 have analyzed which was first officially reported in Wuhan City, in December 2019. Firstly the data have explored in terms of information and quality and after that, the data have cleaned and gone through with feature engineering. Analyzed different types of machine learning-based prediction methods, namely Linear Regression, ARIMA, and SARIMA on the spread of COVID-19 in different regions all over the world. In the end, It has been concluded with the best machine learning model among them for COVID-19 spread forecasting based on theoretical and results in analysis. And also we have discussed that how deep learning can be considered with data limit problem in order to improve the result more dynamically with combination and comparisons of state-of-art approaches for time series problems.

Ensuring the food security of the Russian Federation requires improving the methods of modeling the productivity of agricultural crops, for which various approaches are used in agricultural production, including nonlinear statistical technologies. The article presents the developed method of constructing a mathematical model of time series (TS) for predicting the yield of grain crops in the Statistic system. The forecast is based on representative samples based on the example of two TS—long-term yield levels and statistical data on electricity generation in the Russian Federation. The methodology of mathematical modeling of TS is justified taking into account a priori information about the structure of the simulated TS. The technique involves identifying the cyclical nature of TS at the stage of pre-forecast analysis. For the short-term description of the data, an exponential smoothing model was used. For long-term forecasting, the choice of the seasonal autoregression model and the integrated moving average ARIMA (0,1,1)(0,1,1) is justified, its parameters are determined and its adequacy to real statistical data is proved. Using the found mathematical model, a long—term prediction (period—12 months) was performed on the example of electricity generation in the Russian Federation with a 90% confidence interval. The methodology presented in this paper for analyzing non-stationary time series using computer modeling tools Statistica can also be used to predict retrospective data in various fields of scientific and applied activity for the analysis and forecasting of TS, which are characterized by a seasonal component and a trend.

The World Health Organization (WHO) defines a pandemic as the worldwide spread of a new disease, which stands for coronavirus disease 2019, Covid-19. Since December 2019, the Covid-19 causes infection of over 130 million people and over 3 million deaths by March 3, 2021. In this study, we use fuzzy stochastic differential equations (FSDEs) to describe population dynamics including fuzzyfied stochastic Richards growth model. Due to the lack of information or misleading information about the number of reported cases of Covid-19, we could think that these reported numbers given by authorities might not be very accurate and reliable. In this study, to solve FSDE, the Euler-Maruyama (EM) method which is one of the simplest numerical approximations between other methods for the FSDE is used. Also, fuzzy time series forecasting models have a great scope in the present area especially in the case of epidemic diseases projection. For this reason, Fuzzy ARIMA models are used for forecasting the future growth of Covid-19 in Turkey. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.

Sales forecasting of vegetables and fruits imposes a challenging task for the retailers because the demand for them varies depending on several factors, such as temperature, season, holiday. Poor sales forecasting can cause too much cost for retailers since these products are unusable after deterioration. Also, people tend to consume these products freshly. This research aims to compare the forecasting performance of traditional statistical and new machine learning methods. We apply seasonal ARIMA to forecast daily sales of fruits and vegetables as a traditional method. As a machine learning algorithm, we apply LSTM and XGBoost algorithms. The results indicate that the XGBoost algorithm gives more accurate results than the other two methods.

Temporal Internet of things (IoT) data is ubiquitous. Many highly accurate prediction models have been proposed in this area, such as Long-Short Term Memory (LSTM), Autoregressive Moving Average Model (ARIMA), and Rolling Window Regression. However, all of these models employ the direct-previous window of data or all previous data in the training process; therefore, training data may include various data patterns irrelevant to the current design that will reduce the overall prediction accuracy. In this paper, we propose to look for the previous historical data for a pattern that is close to the current one of the data being processed and then to employ the next window of data in the regression process. Then we used the Support Vector Regression with Radial Basis Function (RBF) kernel to train our model. The proposed model increases the predicted data’s overall accuracy because of the high relevancy between the latest data and the extracted pattern. The implemented methodology is compared to other famous prediction models, such as ARIMA and the rolling window model. Our model outperformed other models with a 9.91 Mean Square Error (MSE) value compared with 12.02, 18.79 for ARIMA and rolling window, respectively.

Gross Domestic Product (GDP) is the value of all the finished goods and services produced within the country in a specific time period. It is a common indicator used to measure a nation’s economic growth. In this paper, the authors used the Box-Jenkins method to build an Auto Regressive Integrated Moving Average (ARIMA) which is suitable for Vietnam’s GDP data. The annual GDP data of Vietnam is collected from Asian Development Bank (ADB) from 1985 to 2019. The appropriate model to forecast Vietnam’s GDP growth rate is ARIMA (3, 1, 3). Finally, the ARIMA model was used to forecast Vietnam’s GDP growth from 2020 to 2025.

Buildings consume about 40% of the world's energy use. Energy efficiency in buildings is an increasing concern for the building owners. A reliable energy use prediction model is crucial for decision-makers. This study proposed a hybrid machine learning model for predicting one-day-ahead time-series electricity use data in buildings. The proposed SAMFOR model combined support vector regression (SVR) and firefly algorithm (FA) with conventional time-series seasonal autoregressive integrated moving average (SARIMA) forecasting model. Large datasets of electricity use in office buildings in Vietnam were used to develop the forecasting model. Results show that the proposed SAMFOR model was more effective than the baselines machine learning models. The proposed model has the lowest errors, which yielded 0.90 kWh in RMSE, 0.96 kWh in MAE, 9.04% in MAPE, 0.904 in R in the test phase. The prediction results provide building managers with useful information to enhance energy-saving solutions.

As the effect of COVID-19 is increasing rapidly every day, it becomes very difficult for the survival of many people and there is a high effect on the economic situations of every country which is affected by it. In the same way, it is affecting all states of India and causing an economic crisis. This paper deals with the analysis of the impact caused by COVID-19 on each state in India and also gives an estimated date on which the effect will reduce along with the analysis report of overall India. For the predicting of the effect, we have used Auto-Regressive Integrated Moving Averages (ARIMA) algorithm which has produced Root Mean Squared Error (RMSE) of around 5.89 for some of the states and other with 20.05 due to the data abnormality. The predicted data for each state projected in the figure using the line plots. And the resulted graphs are explained clearly. The accuracy of the proposed model is around 94.6–96.8% for the states with good data and less RMSE values and 80% for the states with abnormal data and high RMSE values. From the produced results of the proposed methodology, the dates of which the effect of COVID-19 will decrease is calculated for the states which have a high number of cases.

As energy markets become more and more dynamic, the importance of price forecasting has gained a lot of attention over the last few years. Considering also the introduction of new business models and roles, such as Aggregators and energy flexibility traders, in the constantly evolving energy landscape which follows the general opening of the European electricity markets, the need for anticipating energy price trends and flows holds significant business value. On top of that, the exponential renewable energy sources penetration, adds to the challenges introduced to this dynamic scheme of things. Given their volatile and intermittent nature, supply-demand imbalance can reach critical margins, threatening the overall system stability. In the scope of reducing the power imbalances, a forecast for the imbalance volume will be beneficial either from the perspective of the system operator that could minimise mitigation costs, or the market participants that could target extreme prices for maximising their profit, while effectively managing their risks. The development of a deep learning algorithm for the prediction of the net imbalance volume in the UK market is proposed in this paper in comparison with a common but widely used machine learning approach, namely a gradient boosting trees regression model. The variables which contributed the most on those models were mainly the historical values of net imbalance volume. The deep neural network returns a Root mean squared error (RMSE) and Mean Absolute Error (MAE) equal to 200 and 152 MWh in a range of values between [−1.5, 2.0] GWh, respectively, the gradient boosting trees model has an RMSE and MAE equal to 203 and 154 MWh, in contrast to an ARIMA model that has RMSE and MAE equal to 226 and 173 MWh.

COVID-19 has been the most notorious pandemic affecting the entire world resulting in numerous deaths thus crippling the world economy. While vaccines are in the process of being developed for protection, countries are implementing measures such as social distancing to prevent the spread of the virus. Also, there exists a need for developing mathematical models to predict the rate of spread of COVID-19 and quantify its impact on countries such as India. Towards this goal, we developed a realistic COVID-19 dataset consisting of state-wide distribution of number of cases in India from March-July 2020. Further, we conduct exploratory data analysis on the dataset to understand the states and their corresponding growth rates. This enables us to cluster states with exponential and non-exponential growth rates as well as assess the effectiveness of lockdown imposed to curb the spread of virus. Finally, we develop predictive models using Auto-Regressive Integrated Moving Average (ARIMA) and Long Short-Term Memory Networks (LSTM) on time-series data for top-10 affected states in India to predict the rate of spread and validate their accuracy. Finally, our models can be used to guide the development of mechanisms for optimal resource allocation of healthcare systems and response.

Electricity is one of the key role players to build an economy. Electricity consumption and generation can affect the overall policy of the country. This opens an area for some intelligent system that can provide future insights. Intelligent management for electric power consumption requires future electricity power consumption prediction with less error. These predictions provide insights for making further decisions to smooth line the policy and help to grow economy of the country. Future prediction can be categorized into three categories namely (1) long-term, (2) short-term, and (3) mid-term predictions. For our study, we consider mid-term electricity consumption prediction. Dataset is provided by Korea Electric power supply to get insights for metropolitan city like Seoul. Dataset is in time series so we require to analyze dataset with statistical and machine learning models that can support time series dataset. This study provides experimental results from the proposed models. Our proposed models for provided dataset are ARIMA and LSTM, which look promising as RMSE for training is 0.14 and 0.20 RMSE for testing.

The accurate forecasting of electricity demand is an important issue. The objective of this study is the application of two forecasting methods, Box-Jenkin’s autoregressive integrated moving average (ARIMA) and artificial neural network (ANN), to fit a time series data of electricity demand. Three models of ARIMA, AR (1), ARMA (1, 1), and IMA (1, 1), were utilized to fit the demand data. Another method, ANN, was based on the activation function of rectified linear unit (ReLU) and the training algorithm of resilient backpropagation with weight-backtracking. The performance of these models was compared by considering the root mean square error (RMSE) which was used to measure the errors when each model was applied to fit the data.

Building a system to forecast Covid-19 infected cases is of great importance at the present time, so in this article, we present two systems to forecast cumulative Covid-19 infected cases. The first system (DLM-System) is based on deep learning models, which include both long short-term memory (LSTM), bidirectional long short-term memory (Bi-LSTM), and Gated recurrent unit (GRU). The second system is a (TS-System) based on time series models and neural networks, with a Prioritizer for models and weights for time series models acting as an ensemble between them. We did a comparison between them in order to choose the best system to forecast cumulative Covid-19 infected cases, using the example of 7 countries. As some of them have finished the second wave and others have finished the third wave of infections (Russia, the United States of America, France, Poland, Turkey, Italy, and Spain). The criterion for choosing the best model is MAPE. It is a percentage, not an absolute value. It was concluded that an ensemble method gave the smallest errors compared to the errors of the models in the (TS-System).

Wind speed has an important impact on the navigation of ships at sea. If the wind speed can be accurately predicted, the safety of ship navigation would be greatly improved. This paper proposes a wind speed series prediction model based on SARIMA and LSTM. Firstly, the SARIMA model is used to predict and model the observed wind speed sequence data to obtain the predicted value and the residual between the predicted value and the observed value. Training the long and short-term memory neural network with the residual sample set to get a trained network for residual prediction. Finally, to sum the two parts predicted values up to obtain the predicted value of the wind speed series. In order to test the prediction effect of this model, a deep learning environment based on Keras was built, and 5 days of real-time wind speed data in a certain sea area of the South China Sea was used as the input of the model. The prediction results are compared with the prediction results of SARIMA model, LSTM network model, BP network model, LSTM and ARIMA combined model. The experimental results show that the model has high accuracy and less error in the prediction of wind speed series.

This study aims to assess the agricultural drought risk for the period of 1960–2011 and to identify the sustainable adaptation measures in the western part of Bangladesh which is most drought-prone areas in the country. The MK test, Sen’s slope estimator, and ARIMA model have been applied to climatic variables for detecting the trends and forecasting future climate scenarios. The Markov chain analysis and Drought Vulnerability Index (DVI) have been used to generate the Drought Index (DI) and Drought Hazard Index (DHI). The pattern of drought risk is mapped by multiplying hazard and susceptibility indices. The annual average temperature in the area is about 25.44 °C, and the annual, as well as seasonal mean temperature is the lowest in north-east and the highest in the south-western part. The predicted annual mean temperatures vary from 24.47 to 26.75 °C and are higher than long term observed mean although the predicted spatial pattern will remain the same in the area. Drought poses a great risk in agriculture in the northern districts and comparatively low in the southern districts (coastal areas) and 21.64%, 26.53% and 29.67% of the area pose a very high, high, and modest risk respectively. Hence, it is urgent to act on adaptation measures in response to future climate change issues such as raising temperature and rainfall variability.

Traffic congestion research is on the rise, thanks to urbanization, economic growth, and industrialization. Developed countries invest a lot of research money in collecting traffic data using Radio Frequency Identification (RFID), loop detectors, speed sensors, high-end traffic light, and GPS. However, these processes are expensive, infeasible, and non-scalable for developing countries with numerous non-motorized vehicles, proliferated ride-sharing services, and frequent pedestrians. This paper proposes a novel approach to collect traffic data from Google Map’s traffic layer with minimal cost. We have implemented widely used models such as Historical Averages (HA), Support Vector Regression (SVR), Support Vector Regression with Graph (SVR-Graph), Auto-Regressive Integrated Moving Average (ARIMA) to show the efficacy of the collected traffic data in forecasting future congestion. We show that even with these simple models, we could predict the traffic congestion ahead of time. We also demonstrate that the traffic patterns are significantly different between weekdays and weekends.

Based on the monthly social fixed asset investment data of Shanghai from January 2003 to February 2020, this paper uses the SARIMA model and the X_12_ARIMA seasonal adjustment method to make predictions. The predicted value is compared with the actual value. Based on relative error, the predicted value based on the SARIMA model is more accurate and reasonable than the predicted value based on the X-12-ARIMA seasonal adjustment method. SARIMA (2, 1, 1) (0, 1, 1)12 can be used for future forecasts and can provide a basis for future fixed asset investment in Shanghai.

Background: COVID-19 has posed the greatest threat to the world in terms of health and economy. Countries all over the world have acquired different measures to contain COVID-19 spread. Present study focuses on compilation of measures taken by Government of India to fight COVID-19 in major affected states of India.

Artificial neural networks (ANNs) are flexible computing frameworks and universal approximates that can be applied to a wide range of time series forecasting problems with a high degree of accuracy. However, despite all advantages cited for artificial neural networks, their performance for some real time series is not satisfactory. Improving forecasting, especially time series forecasting accuracy is an important, yet often difficult task facing forecasters. The purpose of this paper is to use artificial neural networks and traditional methods (ARIMA model) to forecasting time series, and to diagnose the best method for prediction. The research sample included data of China’s crude oil production chain for the period 1980–2015. The most important results that were reached through this paper are: The results proved that the best method for predicting time series is artificial neural networks, Wherever the error results improved by a large margin it was 99.5%.

The consumption of drinking water in the populations in the last 20 years has varied for different reasons, thus, it is necessary to determine its behavior for sustainable use of the resource. The objective of this work was to present the variation in drinking water consumption during the Sars-Cov-2 health emergency through dynamic modeling in the city of Macas in the Amazon region of Ecuador. The integrated moving average autoregressive model was used for the study (ARIMA), predicting the behavior of drinking water consumption for the year 2021 in the different neighborhoods of the city and relating this result with socioeconomic variables. The prediction for the year 2021 presented a decrease of 0.21% in the volume consumed compared to 2020, in April 2021 an increase of 85.21% was observed compared to the consumption of 2019, which can be attributed to the effects of the pandemic. The highest water consumption occurred in the cluster of neighborhoods with the highest population density, medium-high socioeconomic status, and high availability of basic services. The study aims to provide a valid alternative for decision-making in the framework of a health crisis, as well as possible conflicts in vulnerable areas in the face of the pandemic that affects the entire world.

Time-series data has generally been difficult to predict, due to unseen hidden patterns in the data and the unpredictability of values in them. Classical methods such as autoregressive moving average (ARMA), autoregressive integrated moving average (ARIMA), and artificial neural networks (ANNs) have been used to obtain predictions. However, the accuracy obtained was distorted due to significant misprediction rates. Deep learning, a part of machine learning, has recently seen a surge in usage to obtain accurate predictions from time-series data. Areas such as stock market, petroleum production, solar energy production, electric load, etc. make use of deep learning. Recurrent neural networks are used in most cases due to their ability to recall past information and apply it for future predictions. Variations of classic RNNs such as long short-term memory (LSTM), gated recurrent unit (GRU), and modifications of the previous two networks with other types of networks have been observed. A survey of usage of variations of RNNs and their modified versions has been conducted to understand the strengths of each type of RNN and to find out which type of RNN is the most versatile.

Two approaches are most frequently used to predict the development of cryptocurrency market: deterministic and stochastic. The deterministic approach seeks to explain the development of cryptocurrencies through the relationship of several indicators. A stochastic approach (such as ARIMA) seeks to optimize the parameters of a statistical model. This article aims to compare approaches to the assessing cryptocurrencies development using the number of active Bitcoin and Ethereum wallets. For this purpose, a deterministic model based on the Verhulst equation, and a stochastic model based on ARIMA was formulated. The results show that the usage of relative differences wins over absolute ones. At the same time, the predictive value of a purely deterministic model on short segments is not very high, but it has the advantage in analytical form. Further will focus on the combination of a deterministic Bass-type model and statistical methods with stochastic analysis tools.

This paper presents a comparative analysis of predictive models applied to air quality estimation. Currently, among other global issues, there is a high concern about air pollution, for this reason, there are several air quality indicators, with carbon monoxide (CO), sulfur dioxide (SO2), nitrogen dioxide (NO2) and ozone (O3) being the main ones. When the concentration level of an indicator exceeds an established air quality safety threshold, it is considered harmful to human health, therefore, in cities like London, there are monitoring systems for air pollutants. This study aims to compare the efficiency of stochastic and heuristic predictive models for forecasting ozone (O3) concentration to estimate London's air quality by analyzing an open dataset retrieved from the London Datastore portal. Models based on data analysis have been widely used in air quality forecasting. This paper develops four predictive models (autoregressive integrated moving average - ARIMA, support vector regression - SVR, neural networks (specifically, long-short term memory - LSTM) and Facebook Prophet). Experimentally, ARIMA models and LSTM are proved to reach the highest accuracy in predicting the concentration of air pollutants among the considered models. As a result, the comparative analysis of the loss function (root-mean-square error) reveled that ARIMA and LSTM are the most suitable, accomplishing a low error rate of 0.18 and 0.20, respectively.

In recent years, the smart city concept has become a very actual topic of many scientific publications and the implementation of many technical and IT solutions. The smart city is a city that uses information and communication technologies to increase the interactivity and efficiency of urban infrastructure and its components, as well as to raise the awareness of its inhabitants. The article presents Smart City’s concepts and the prospects for their development in the context of the growing population of cities in the world. Nowadays cities are forced to combine various aspects in order to provide a high quality of life, comfort and a friendly environment. These aspects include areas related to the economy, environment, management and mobility. Using ARIMA - Autoregressive Integrated Moving Average and Exponential Smoothing algorithms, forecasts of growth of selected parameters from current solutions were carried out. In particular, the analysis and prediction of traffic were concentrated on the example of a selected city of Rzeszow in Poland, which has been a leader in the implementation of Smart City services for many years. Based on historical data, the correctness of predictions was assessed. Moreover, the directions and possibilities of development of smart cities as well as production organization in Industry 4.0 were determined.

In order to solve the problem that oxygen saturation can be described during continuous monitoring. We use ARIMA model to describe the change of blood oxygen saturation of each person, and use pca-ga-bp neural network model to represent a person by inputting parameters under the new index system. The network topology is constructed by 33 training samples, and the index of the three test samples is the quantitative value of the sample number of the three test samples. The accuracy of the test model is 100%. Compared with BP neural network system, pca-ga-bp neural network model can more accurately represent a person. Then, the correlation between age and SpO2 was analyzed by data visualization, deterministic coefficient method and regression coefficient. First of all, we add the mean value and variance of 36 groups of SpO2 data to describe the average level and stability of individual SpO2, and correlate them with age, smoking history, gender and BMI to form a series model of SpO2. We get: the correlation between age and mean, variance and BMI is very small, but also affected; age and mean is negatively correlated, age and variance is positively correlated, age and BMI is positively correlated. In other words, compared with young people, the average oxygen saturation of the elderly is lower, the stability of individual oxygen saturation is lower, and they are more likely to gain weight.

The COVID-19 now became a pandemic and rising rapidly and spreading in all parts of the world like fire. India reported its first COVID-19 case on January 30, when a student arrived in Kerala from Wuhan. Thousands of people are acquiring this deadly virus daily and with many people dying from it. The major concern of all the countries is to protect its citizens and try to eradicate this disease as fast as possible. This paper aims to perform exploratory analysis using the concepts of data science on the confirmed cases, total deaths, and total recovered cases of this virus. The research work predicts the spread of the outbreak for the next five days by using time-series forecasting algorithms. This paper deals with learning about how the corona virus is spreading and using that trend to predict for the upcoming days. It would be able to predict to a suitable accuracy which can help the government learn about the statistics of this disease and prepare further for protection against this. The results are discussed at last with prediction and error estimates.

Forecasting the flow of rivers is essential for maintaining social well-being since their waters provide water and energy resources and cause serious tragedies such as floods and droughts. In this way, predicting long-term flow at measuring stations in a watershed with reasonable accuracy contributes to solving a range of problems that affect society and resource management. The present work proposes the MultiTask-LSTM model that combines the recurring model of Deep Learning LSTM with the transfer of learning MultiTask Learning, to predict and share information acquired along the hydrographic basin of Paraiba do Sul river. This method is robust for missing and noisy data, which are common problems in inflow time series. In the present work, we applied all 45 measurement stations’ series located along the Paraiba do Sul River basin in the MultiTask-LSTM model for forecasting the set of these 45 series, combining each time series’s learning in a single model. To confirm the MultiTask-LSTM model’s robustness, we compared its predictions’ results with the results obtained by the LSTM models applied to each isolated series, given that the LSTM presents good time series forecast results in the literature. In order to deal with missing data, we used techniques to impute missing data across all series to predict the 45 series of measurement stations alone with LSTM models. The experiments use three different forms of missing data imputation: the series’ median, the ARIMA method, and the average of the months’ days. We used these same series with imputing data in the MultiTask-LSTM model to make the comparison. This paper achieved better forecast results showing that MultiTask-LSTM is a robust model to missing and noisy data.

The presence of soil moisture is of paramount importance in agricultural domain, as it constitutes largely towards variations in soil texture, and development of crops. Hence, evaluation of this parameter can turn out to be very effective while performing agricultural activities. Mainly, evaluations for the parameter is done with an aim to estimate and reduce water consumption within the fields. In this article, the presence of soil moisture has been evaluated at three different levels, i.e., 10 cm, 45 cm, and 80 cm through Autoregressive Integrated Moving Average (ARIMA) modeling technique based on Time Series Analysis, to predict the future possible values so that precise distribution water can be done within the fields. The intermediate diminution in error rates attained by the modeling technique attained between 74%–77% in comparison to other modeling techniques, depicting its superiority. Based on the results, distribution of water was scheduled in advance as per the minimal requirement, resulting lesser consumption and better crop yields

The world is currently facing a Covid-19 pandemic and that virus is spreading rapidly among people, which leads to an increase in the number of infection cases and also an increase in the number of death cases. This is a huge challenge as this pandemic affected all sectors, and therefore there was important for mathematicians in modelling this epidemic spread in the world to reduce the damage caused by this pandemic and also discovering the pattern of that virus spreading. In our report, time series models are used to obtain estimates of the number of cases of infection and numbers of deaths using ARIMA, Holt’s Linear Trend, BATS, TBATS, and SIR Models. We have developed a new algorithm to use these models and choose the best model for forecasting the number of infections and deaths in terms of the least error of MAPE as standard. We have observed in most of the data that were used in this algorithm that the best models that achieve the least forecast errors are BATS, TBATS, and ARIMA respectively. The experiment was held for the ten countries most affected by the Covid-19, this algorithm was able to detect the data pattern of the virus spreading for every country, besides it is interested in more research and studies on other models.

The investigation of economic sides of Greenhouse Gas (GHG) emissions and its penalties is very important, particularly in terms of its volume at the present increasing trend. Carbon dioxide (CO2) emissions account for the largest proportion of total greenhouse gas emissions produced mainly by human activities. So, the prediction of air pollution due to emissions of CO2 can give the right way to policies accepted. In the economics literature of the last decades, the relationship between emissions of CO2 and financial progress is of great interest. This study aimed at modeling and forecasting some environmental and economic variables and investigating the existence of long-run equilibrium relationship between major GHG emissions and economic growth in eight SAARC countries—Bangladesh, India, Pakistan, Sri Lanka, Nepal, Bhutan, Maldives, and Afghanistan. Time series data from 1990 to 2018 was collected from World Bank data archive. Autoregressive Integrated Moving Average (ARIMA) Models and cointegration theory was applied to analyze the data. While forest areas in most of the countries showed decreasing trend, GHG and CO2 emission showed increasing trend in majority of the countries. Industrial and GDP growth in the region was slowly growing over time. ARIMA models fitted well to the data except the cases where data did not show any variability. Mix results were obtained regarding the existence of coingration between CO2 emission and industrial growth and that between CO2 emission and GDP growth. These findings would enable the environmental authorities to understand the environmental impacts of economic development on degradation and to use time series approaches to handle environmental problems.

Identifying optimal models to forecast economic cycles has been a point of great consideration in literate. A key point of debate in the literature is whether linear or non-linear models perform best at forecasting economic cycles. The literature largely forces on the forecasting of business cycles, and very limited work has been done on financial cycle forecasting. Given the proven destructiveness of financial cycles, the ability to accurately forecast future financial cycle movements in an economy could aid policymakers in managing such cycles. This article evaluates the forecasting performance of both the non-linear Markov Regime-Switching Autoregressive methodology and Smooth Transition Autoregressive methodology relative to the benchmark ARIMA model in forecasting the aggregate South African financial cycle over different time horizons. A fixed window rolling forecast approach is followed, whereby the performance of forecasting the aggregate South African financial cycle 3-steps forward, 6-steps forward, 12-steps forward, 18-steps forward and 24-steps forward is evaluated. The findings indicate that the linear ARIMA model outperforms the non-linear MSMV-AR and LSTAR models at forecasting short periods ahead such as 3–6 months ahead. However, both the MSMV-AR and LSTAR models outperform the ARIMA model, given a longer time horizon such as 12–24 months. Hence, to forecast the aggregate South African financial cycle 3–6 months ahead policymakers should use an ARIMA. However, the MSMV-AR and LSTAR models should be used to forecast the aggregate South African financial cycle 12–24 months ahead.

Early prediction of the crime occurrence reduces its impact. Several studies have been conducted to forecast crimes. However, these studies are not highly accurate, particularly in short-term forecasting such as over one week. To respond to this, we examine sequence to sequence (Seq2Seq) based encoder-decoder LSTM model using two real-world crime datasets of Brisbane and Chicago, extracted from the open data portal, to make one week ahead of total daily crime forecasting. We have built an ARIMA statistical model and three machine learning-based regression models that differ in their architecture, namely, simple RNN, LSTM, and Conv1D with a novel approach of walk-forward validation. Using a grid search strategy, the hyperparameters of the models are optimized. The obtained results demonstrate that the proposed Seq2Seq model is highly effective, if not superior, compared to its counterparts and other algorithms. This proposed model achieves state-of-the-art results with a relatively Root Mean Squared Error (RMSE) of 0.43 and 0.86 on both datasets, respectively.

Low latency is an important index in LEO satellite communication systems, while the satellite capacity “application-assignment” scheme based on the DVB-RCS2 standard in the return channel causes a long round-trip delay. To improving the quality of experience (QoE), in this paper, a more aggressive capacity pre-assignment scheme combining traffic prediction and free capacity assignment (FCA) is proposed. The network control center (NCC) predicts traffic for every return channel satellite terminal (RCST) and assigns capacities in advance without capacity requests. Several FCA strategies based on multi-frequency time division multiple accesses (MF-TDMA) in the physical layer are analyzed as a compensatory capacity assignment method to deal with the inaccuracy of traffic prediction. Simulation results show that the proposed FCA strategies have better performance than existing FCA strategies.

This paper develops a framework for the analysis of state-space models combined with Kalman and smoothed Kalman filters for the estimation of unknown states, and parameters, determining the accuracy of the algorithms, with the purpose of analyzing some time series of the macroeconomy of Ecuador. This methodology plays an important role in the area of economics and finance and has many advantages because it allows describing how observed macroeconomic variables can be related to potentially unobserved state variables, determining the evolution in real time, estimating unobserved trends, changes of structures and make forecasts in future times. To achieve the objectives, three models are proposed: the first model is used to estimate the Ecuador’s gross domestic product. The second model combines a state space model with the classic ARIMA (p, q, r) model to adjust the GDP rate and finally it is considered a model for the simultaneous stress time series analysis related to: consumer price index, industrial production index and active interest rate. In all the cases studied, the estimates obtained reflect the real behavior of the Ecuadorian economy. The square root of the mean square error was used as a measure of goodness of fit to measure the quality of estimation of the algorithms, obtaining small errors.

COVID-19 is a real problem, and it is spreading like a forest fire. The data of this pandemic is time-series data. The models that can handle time-series data are the ARIMA model, the Holt-Winter model, the SARIMAX model, polynomial regression, and LSTM. These models have been applied to COVID-19 data, and the results are discussed with significance. This chapter used three types of datasets. The primary dataset is the 2019 Novel Coronavirus COVID-19 (2019-nCoV) Data Repository by Johns Hopkins CSSE (https://github.com/CSSEGISandData/COVID-19). The second dataset is used from Worldometers website (https://www.worldometers.info/), and third is from Kaggle. The SARIMAX model produced 0.236 as the MAPE value, while the Holt-Winter model produced 0.249. The polynomial regression model shows that the accuracy of the model approximated for the tenth day is 85% in the prediction of the number of affected cases and the number of deaths. The LSTM model used the ADAM optimizer and calculated the root mean square error. The prediction error for training is 6.45, and the calculated overall error is 5.34.