Time Series Analysis Model Research Articles

ChebyshevNet: a novel time series analysis model using Chebyshev polynomial

ChebyshevNet: a novel time series analysis model using Chebyshev polynomial

Machine learning for predicting tourist spots’ preference and analysing future tourism trends in Bangladesh

ABSTRACT This study uses machine learning, including Support Vector Machines, Decision Trees, K-Nearest Neighbors, to examine Bangladesh’s tourism industry to forecast traveller preferences. We use time series analysis, including ARIMA, Moving Average, and Auto-regression models, to predict future tourism trends. Our results show that, with an accuracy of 96.3%, Linear SVM was the best at predicting preferences. For trend forecasting, the ARIMA model fared better than the others, suggesting that Bangladeshi tourism may be headed in an undesirable direction. Our observations and insights can help guide strategic choices and decisions in the creation of policies and the administration of tourism.

Forecast of consumption of natural gas in U.S. based on time series analysis and ARIMA model

Natural gas is a widely used fossil fuel, playing a vital role in both industrial and economic sectors. As a widely used fossil fuel, the prediction of natural gas consumption is crucial for industrial and economics and hence many scholars have been researched on the topic. Due to the weather variation and the sensitivity of industrial demand, a precise consumption forecast is difficult. In the article, ARIMA model, a widely used model in time series analysis and forecasts would be considered in the prediction for consumption for future one year. It turns out that, in the next year, the consumption of natural gas in the coming year is stable and will stay at a still level and the overall trend of the series is increasing, with no significant short-term fluctuations. Therefore, the article suggested that long-term investment in industries related to natural gas is recommended. Besides, short-term investment is not suggested compared to the long-term due to the still level of the consumption within one year.

Research on UK Unemployment Rate Forecast Based on ARIMA Model

The objective of this study is to predict the unemployment rate in the UK and the Eurozone for the years 2014 to 2024 using the time series analysis model known as ARIMA. It is essential to be informed of changes to employment rates in order to address them and changes in the economic environment including historical occurrences such as Brexit, COVID-19, and shifts in the labor market. The ARIMA model was used because it fits non-stationary time series, which is a feature present in the employment rates of both time series. The monthly unemployment rates for each state were seasonally adjusted and the ARIMA parameters were estimated by the automatic method. The study observed that while the interior portion of the trading cycle was effectively predicted by the ARIMA model, it failed to do so for the short-term UK unemployment in the post-2023 period because of exogenous shocks. However, the model was equally successful in predicting a marginal contraction in unemployment rates in the Euro zone after 2023. By performing the Ljung-Box test, it was established that the residuals of the models were random and therefore, the models were fit. Accordingly, the results suggest that the estimates derived from the ARIMA model are beneficial in economic predictions, which helps governments to design better social security programs and other economic plans to address the changing employment landscape.

A tale of two times: an exploration of healthcare utilization patterns before and during COVID-19 in Iran

BackgroundThe COVID-19 pandemic has impacted global healthcare utilization patterns. This study aimed to examine the impact of COVID-19 pandemic on utilization rate of healthcare services in Iran.MethodIn this quasi-experimental study, data on the utilization rates of laboratory services, sonography exams, CT scans, MRIs, and EKGs was collected from the Social Security Organization (SSO)’s insurance information system. This data, covering 47 months prior to the pandemic and 25 months during it, was analyzed to assess the pandemic’s impact on healthcare utilization among insured individuals in Iran. The data was categorized into direct, indirect, and total sectors, and an Interrupted Time Series Analysis (ITSA) model was employed for data analysis, examining both total and sector-specific utilization rates.Findings: The study for single group indicated that in the total sector, Utilization rate per 1000 insured significantly decreased by 25.25 for laboratory services, decreased by 3.99 for sonography, decreased by 1.08 for MRIs and decreased by 1.01 for EKGs, but increased by 2.28 for CT scans in the first pandemic month. Over following months, monthly utilization trends per 1000 insured increased significantly- laboratory services + 1.08, sonography + 0.11, CT scans + 0.12, MRIs + 0.06, and EKGs + 0.05. Pre-pandemic, monthly utilization per 1000 insured was 62.68 labs, 14.47 sonography, 0.72 CT scans, 2.06 MRIs, with all significantly higher in the indirect over direct sector except EKGs which were 2.08 higher in the direct sector. In the first pandemic month, there were significant between-sector differences per 1000 of -4.4 for sonography, + 1.89 CT scans, -1.01 MRIs and + 1.29 EKGs.ConclusionThe COVID-19 pandemic led to a significant decline in healthcare service utilization, particularly in total and direct sectors, while CT scans remained unaffected. To address these challenges and meet patient needs, Iran’s health system should adopt alternative delivery methods like telemedicine.

The Relationship Between the Development of the Art Industry and the Turnover of Economic Cycles

Abstract: In recent years, the area has grown significantly, creating a trend in art appreciation. This boom has led to a surge in the popularity of museums such as the National Gallery. This project aims to find out more about the economic contribution of art exhibitions. Using advanced time series analysis models including the intervention analysis model and vector autoregressive model, we obtained the following conclusions. First, the positive impact of low levels of economic development and high levels of government expenditure on the arts industry is obvious. Second, the growth of the arts industry has a small impact on most economic indicators. Third, the final impact of GDP on the arts industry is still uncertain due to the complex pattern of the model, with the possibility of a positive correlation with ln (art prices) being observed in the early stages of small GDP growth, while a quadratic pattern may be observed when the GDP index grows above a certain level and art prices fall Conclusion. Further research is needed to investigate the complex relationship between inflation and art prices. Our academic contribution mainly focuses on the multiple-way causal relationships between the economy, art exhibition, tourism, and education, which reveals the underlying mechanism of how art exhibitions promote the economy.

Recent achievements in nonlinear dynamics, synchronization, and networks.

This Focus Issue covers recent developments in the broad areas of nonlinear dynamics, synchronization, and emergent behavior in dynamical networks. It targets current progress on issues such as time series analysis and data-driven modeling from real data such as climate, brain, and social dynamics. Predicting and detecting early warning signals of extreme climate conditions, epileptic seizures, or other catastrophic conditions are the primary tasks from real or experimental data. Exploring machine-based learning from real data for the purpose of modeling and prediction is an emerging area. Application of the evolutionary game theory in biological systems (eco-evolutionary game theory) is a developing direction for future research for the purpose of understanding the interactions between species. Recent progress of research on bifurcations, time series analysis, control, and time-delay systems is also discussed.

Economic Development and the Development of the Securities Market: A Study of China's CSI 300 Index

The Chinese securities market has experienced over two decades of development and has emerged as a significant global investment market. However, compared to other developed securities markets, it still faces issues such as an inadequate institutional framework, irrational investor behavior, and imperfect mechanisms. To explore the relationship between China's economic development and its securities market, this paper employs Eviews time series analysis and VAR modeling for empirical research, aiming to uncover the relationship and operational mechanisms between the two. The findings indicate a certain correlation between economic development and securities market development in China. The variance decomposition reveals that the contribution of economic development to the securities market is relatively low, highlighting a major issue within the Chinese securities market. Consequently, this paper proposes reform suggestions from various perspectives, including institutional, investor, and listed company aspects, to guide the healthy growth of the Chinese securities market and create a favorable environment for economic development.

Pollutants-mediated viral hepatitis in different types: assessment of different algorithms and time series models

The escalating frequency of environmental pollution incidents has raised significant concerns regarding the potential health impacts of pollutant fluctuations. Consequently, a comprehensive study on the role of pollutants in the prevalence of viral hepatitis is indispensable for the advancement of innovative prevention strategies. Monthly incidence rates of viral hepatitis from 2005 to 2020 were sourced from the Chinese Center for Disease Control and Prevention Infectious Disease Surveillance Information System. Pollution data spanning 2014–2020 were obtained from the National Oceanic and Atmospheric Administration (NOAA), encompassing pollutants such as CO, NO2, and O3. Time series analysis models, including seasonal auto-regressive integrated moving average (SARIMA), Holt-Winters model, and Generalized Additive Model (GAM), were employed to explore prediction and synergistic effects related to viral hepatitis. Spearman correlation analysis was utilized to identify pollutants suitable for inclusion in these models. Concurrently, machine learning (ML) algorithms were leveraged to refine the prediction of environmental pollutant levels. Finally, a weighted quantile sum (WQS) regression framework was developed to evaluate the singular and combined impacts of pollutants on viral hepatitis cases across different demographics, age groups, and environmental strata. The incidence of viral hepatitis in Beijing exhibited a declining trend, primarily characterized by HBV and HCV types. In predicting hepatitis prevalence trends, the Holt-Winters additive seasonal model outperformed the SARIMA multiplicative model ((1,1,0) (2,1,0) [12]). In the prediction of environmental pollutants, the SVM model demonstrated superior performance over the GPR model, particularly with Polynomial and Besseldot kernel functions. The combined pollutant risk effect on viral hepatitis was quantified as βWQS (95% CI) = 0.066 (0.018, 0.114). Among different groups, PM2.5 emerged as the most sensitive risk factor, notably impacting patients with HCV and HEV, as well as individuals aged 35–64. CO predominantly affected HAV patients, showing a risk effect of βWQS (95% CI) = − 0.0355 (− 0.0695, − 0.0016). Lower levels of PM2.5 and PM10 were associated with heightened risk of viral hepatitis incidence with a lag of five months, whereas elevated levels of PM2.5 (100–120 μg/m3) and CO correlated with increased hepatitis incidence risk with a lag of six months. The Holt-Winters model outperformed the SARIMA model in predicting the incidence of viral hepatitis. Among machine learning algorithms, SVM and GPR models demonstrated superior performance for analyzing pollutant data. Patients infected with HAV and HEV were primarily influenced by PM10 and CO, whereas SO2 and PM2.5 significantly impacted others. Individuals aged 35–64 years appeared particularly susceptible to these pollutants. Mixed pollutant exposures were found to affect the development of viral hepatitis with a notable lag of 5–6 months. These findings underscore the importance of long-term monitoring of pollutants in relation to viral hepatitis incidence.

Relationship of litterfall anomalies with climatic anomalies in a mangrove swamp of the Yucatan Peninsula, Mexico.

Among the set of phenological traits featuring mangrove ecosystems, litterfall production stands out with marked intra-annual and longer-term variation. Furthermore, mangrove forests resilience is one of the most important ecological attribute, reconciling the juxtaposed terrestrial and marine environment such transitional systems occupy. However, world's mangroves are nowadays facing recurrent climatic events, reflected in anomalies depicted by major drivers, including temperature and precipitation. This physical-environmental setting may either constrain or favor overall forest productivity. A combination of time series analysis (spectral density and cross-correlation techniques) and statistical model fitting (General additive model) was implemented to explore trends in total litterfall of a well-developed mangrove forest in southeastern Gulf of Mexico (Celestun Lagoon, SE Mexico) and potential association with the varying behavior of temperature (°C) and precipitation (mm month-1), highlighting their anomalies. The results are consistent with a synchronous response between litterfall production and climatic variables (mean monthly temperature and total monthly precipitation). Concurrent peak litterfall production in Celestun lagoon with high temperatures and precipitation occurred during June and October, featuring a two-month time lag for the response time. More than half of the litterfall anomalies (53.5%) could be reflecting either multiple sources of climatic anomalies (maximum, minimum, and monthly average temperature and monthly total precipitation) or single point events (cyclone landfall). This relationship dynamics showed an interannual persistence (1999-2010). The structure portrayed by the litterfall time-series was not unequivocally related to climatic anomalies. Arguably, climatic anomalies behave with different intensities and even may exhibit complex interactions among them. The study of anomalies provides a baseline for a better grasp of: i) mangrove anomalies responses and ii) their vulnerability to these extremes.

Development, Calibration and Validation of Time Series Analysis and Artificial Neural Network Joint Model for Urban Noise Prediction

Noise in large urban areas, which is mainly generated by road traffic and by the human activities carried out nearby and inside the area under study, is a relevant problem. The continuous exposure to high noise levels, in fact, can lead to several problems, largely documented in the scientific literature. The analysis and forecasting of the noise level in a given area are, then, fundamental for control and prevention, especially when field measurements present peculiar trends and slopes, which can be modeled with a Time Series Analysis approach. In this paper, a hybrid model is presented for the analysis and the forecasting of noise time series in urban areas: this technique is based on the application of a deterministic decomposition model followed in cascade by a predictor of the forecasting errors based on an artificial neural network. Two variants of the hybrid model have been implemented and presented. The time series used to calibrate and validate the model is composed of sound pressure level measurements detected on a busy road near the commercial port of an Italian city. The proposed hybrid model has been calibrated on a part of the entire time series and validated on the remaining part. Residuals and error analysis, together with a detailed statistical description of the simulated noise levels and error metrics describe in detail the method’s performances and its limitations.

Performance of Deep Learning Models on Imputed Time Series Data: A Simulation Study and Application to Leading Airline Companies' Stock Price

In this study, the validity of imputation techniques for deep learning methods in time series analysis is investigated using datasets based on daily closing data in the stock market. Datasets of daily closing stock prices for Turkish Airlines, Deutsche Lufthansa AG, and Delta Airlines, as well as a simulated dataset, are used. LSTM, GRU, RNN, and Transformer models, which are deep learning models, are employed. The original dataset and datasets with 5%, 15% and 25% missing data are analyzed imputing Linear, Spline, Stineman, Mean and Random imputation techniques. The results show that model performance varies depending on the imputation technique and the rate of missing data. GRU and Transformer models are favored for their robustness and excellent performance. For handling missing data, using spline and Stineman imputations is advisable to maintain high model accuracy. This study emphasizes the usability of various imputation techniques and deep learning models in time series analysis. It assesses model performance using both MAPE and RMSE to gain a comprehensive understanding of predictive accuracy and reliability, aiming to guide future research by comparing these methods.

Time Series Analysis for Air Pollution Prediction in High-Intensity Development Areas using Deep Learning

Objective: The purpose of this study is to design novel approach to predict air pollution indices accurately using hybrid approach in highly populated cities of the world. Methods: This study aimed to address the significant bottleneck of relying on a single model for time series analysis which we accomplished by developing a hybrid forecasting method. This blended method combines the advantages of CNN (Convolutional Neural Networks) and LSTM (Long Short Term Memory) hybrid model together, overcoming insufficiency of air pollution prediction. Findings: Due to the empirical operation that revealed in achieving proper predictions we have come to understand that relying on a single model for the estimation of time-series data results is inefficient and misleading, mostly due to the non-linear nature of raw data. On the other hand, by using this hybrid model forecasting, we can uncover more detailed patterns through longer period of time and predict them more accurate by using weather and air pollution related data.The hybrid CNN-LSTM Model achieved least values of 6.742 for MAE,12.921 for RMSE compared to other approaches, which indicatesmere perfection. Novelty: The novelty of our approach is to fuse LSTM and CNN models in order to perform exhaustive analysis on the dynamics of the air pollution throughout the entirety of the city. Through the combination of these two models, such a tool is made simpler to use, and there is an improved accuracy in predictions. This thus becomes a promising tool in the fight against air pollution in cities that have a very high population just like Delhi. Keywords: ­ Air Pollution, LSTM, CNN, Deep Learning, Time Series Analysis, Nonlinear Relationships

Research on the Impact of Economic Development on Financial Derivatives

This paper aims to further investigate the impact of economic growth on the demand for financial derivatives, provide effective support to financial market participants and investors, and promote the stable development of financial markets. Firstly, time series analysis models, including autoregressive, mixed autoregressive, and moving average regression, are established by considering various factors in the financial market. Secondly, the basic characteristics of financial derivatives are thoroughly analyzed, and relevant variables are selected for explanation. Finally, the paper elucidates the fundamental functions of financial derivatives and their role in economic growth. The empirical results show that the correlation coefficient between non-financial institution derivatives and enterprise value is 0.2049, and the impact of enterprise economic growth on derivatives demand is 0.0003. Through time series analysis, the study reveals the impact mechanisms of economic cycles, business fluctuations, and other factors on the financial derivatives market, providing effective support for investors and participants.

On Detecting an Appropriate Model in Time Series Analysis

This study sought to present yet another method of decomposition in time series data. The data for this study were of secondary source and obtained from sources which comprised of both the open and the closed stock prices. The two data were firstly tested for randomness and they were confirmed fit for time series analysis. The two data were also subjected to trend curve analysis, and it was observed that both data were of exponential curve since the exponential trend curve exhibited the highest coefficient of determination (88%), among other trend curves which included linear, quadratic, cubic and logarithmic curves. In the decomposition of the two data series, using the exponential trend, it was revealed that the model, for each data were of multiplicative type since the multiplicative model had the Minimum Mean Squared Error (MSE) of 0.00827 and 0.003665respectively for both Open and Closed Stock Prices of Coca-Cola Data. Hence, in this study, it was recommended that this traditional method of statistics should be applied in the decomposition of any time series data.

A Smart Approach For Budget Deficits PredictionUnder Economic Shocks

Historically, the main behavior of fiscal policy is to distribute resources, income, and expenditures, which are interconnected functions of economic stability. Recently, the scope of public sector economics has expanded beyond budgetary components in parallel with the development of public finance. While the budget reached the economic division form of budget items based on program and performance theories, Budget monitoring, and financial risk management are currently challenging, particularly in the face of monetary policy uncertainty. Financial institutions are crucially concerned with the stability of public finance in low-income countries (LICs) as it contributes to improved investor confidence and fiscal decision-making. Hence, economists investigated uncertainty shocks and contributed to managing financial risks with global and energy uncertainty indices. Furthermore, the maturity of digital transformation and artificial intelligence financial applications catalyzed scholars to examine its contributions in the fiscal distress prediction field. Hence, this research aims to integrate artificial intelligence into financial performance analysis to bridge the gap in budget forecasting. The study was aimed at proposing an Economics Division Uncertainty approach (EDUA), which combined (ARIMA) and (LSTM) models for time series analysis of Nuclear Material Authority expenditures over the previous five years divided into quarterly periods, to achieve efficiency in spending. The (ARIMA) model’s (ADF) results showed that uncertainty indicators are highly significant. The best (p-value) in the first and second differences in (ARIMA) models is (0.0001) for petroleum items, (0.0001) for solar price rates, (0.001) for the US exchange rate, and (0.003) for electricity price rates, when compared to (EDU_LSTM). Both models have similar accuracy rates, with the best being (EDU_ARIMA) (solar price 97%, USD exchange rate 84%). The second study proposed a composite model of four machine-learning tools to enhance financial performance during financial distress. The study collected (12) indicators from general financial literature and corporate studies, utilizing the (XGBOOST, Random Forest, KNN, and Naïve Bayes) models. Comparing the accuracy results for each model presented different accuracy results in the deep learning models over five years of data. The best accuracy score was for Random Forest at (69%), XGBOOT at (68%), and KNN at (68%). We recommended explainable AI as future research to interpret the budget deficit during the fiscal year period. Keywords: Economic Shocks, Uncertainty, Budget Reliability, Financial Distress Prediction, Artificial Intelligence.

Does digitalisation affect the adoption of electric vehicles? New regional-level evidence from Google Trends data

ABSTRACT Digitalisation is an important dimension that contributes to fostering the adoption of electric mobility. We investigate this unexplored topic by focusing on the regional level of analysis, and presenting new data and evidence for a large number of regions in Europe, Canada and the United States. The empirical analysis makes use of Google Trends data. It constructs new indicators of digitalisation and the adoption of electric vehicles, as measured by Google search queries. The new dataset contains indicators for 182 regions in 15 countries for the period 2010–23. We use this dataset to carry out a time-series analysis (vector error correction (VEC) model) of the relations between digitalisation and electric vehicles in each region. The results show that digitalisation is an important factor that has fostered the adoption of electric vehicles in the last decade. The analysis, though, also points out that there is considerable heterogeneity in the time-series results among regions in our sample. Digitalisation has a more visible effect on electric mobility for regions that have higher gross domestic product per capita, better internet infrastructures, a young and well-educated workforce, and higher population density.

Examining sea levels forecasting using autoregressive and prophet models

Global climate change in recent years has resulted in significant changes in sea levels at both global and local scales. Various oceanic and climatic factors play direct and indirect roles in influencing sea level changes, such as temperature, ocean heat, and Greenhouse gases (GHG) emissions. This study examined time series analysis models, specifically Autoregressive Moving Average (ARIMA) and Facebook’s prophet, in forecasting the Global Mean Sea Level (GMSL). Additionally, Vector Autoregressive (VAR) model was utilized to investigate the influence of selected oceanic and climatic factors contributing to sea level rise, including ocean heat, air temperature, and GHG emissions. Moreover, the models were applied to regional sea level data from the Arabian Gulf, which experienced higher fluctuations compared to GMSL. Results showed the capability of autoregressive models in long-term forecasting, while the Prophet model excelled in capturing trends and patterns in the time series over extended periods of time.

Machine Learning Orchestrating the Materials Discovery and Performance Optimization of Redox Flow Battery

AbstractThis review exploits the crucial role of computational methods in discovering and optimizing materials for redox flow batteries (RFBs). Integration of high‐throughput computational screening (HTCS) and machine learning (ML) accelerates materials discovery, guided by algorithms categorizing RFBs. A collaborative exploration, spanning macroscopic to mesoscopic scales, combines quantum machine learning with reinforcement learning, transfer learning, time series analysis, Bayesian optimization, active learning and various generative models. The collaborative integration of ML with computational techniques and experimental methods, anchored in experimentally validated Density Functional Theory (DFT) calculations and molecular dynamics (MD) simulations, proves indispensable for cost‐effective RFBs. Data collection and feature engineering are explored, emphasizing the integration of optimization goals and precise data collection within the ML framework. Feature analysis importance is highlighted, utilizing methods such as the filter, embedded, wrapper and deep learning methods for efficient energy materials exploration. Computational perspectives on materials features and operating conditions encompass membrane characteristics, fluid dynamics, temperature dependence and pressure sensitivity. Time‐dependent features and ML‐generated insights are crucial for understanding cycling performance intricacies, providing a comprehensive understanding of RFB materials.

Endpoint prediction of BOF steelmaking based on state-of-the-art machine learning and deep learning algorithms

Abstract To enhance the efficiency and sustainability, technical preparations were made for eliminating the Temperature, Sample, Oxygen test of basic oxygen furnace (BOF) steelmaking process in this work. Utilizing data from 13,528 heats and state-of-the-art (SOTA) machine learning (ML) and deep learning algorithms, data-driven models with different types of inputs were developed, marking the first use of time series data (off-gas profiles and blowing practice related curves) for BOF steelmaking’s endpoint prediction, and the tabular features were expanded to 45. The prediction targets are molten steel’s concentrations of phosphorus (Endpoint [P], %) and carbon (Endpoint [C], %), and temperature (Endpoint-Temp, °C). The optimal models for each target were implemented at a Hesteel Group’s BOF steelmaking facility. Initially, SOTA ML models (XGBoost, LightGBM, Catboost, TabNet) were employed to predict Endpoint [P]/[C]/Temp with tabular data. The best mean absolute errors (MAE) achieved were 2.276 × 10−3% (Catboost), 6.916 × 10−3% (Catboost), and 7.955°C (LightGBM), respectively, which surpassed the conventional models’ performance. The prediction MAEs of the conventional models with the same inputs for Endpoint [P]/[C]/Temp were 3.158 × 10−3%, 7.534 × 10−3%, and 9.150°C (Back Propagation neural network) and 2.710 × 10−3%, 7.316 × 10−3%, and 8.310°C (Support Vector Regression). Subsequently, predictions were explored to be made using SOTA time series analysis models (1D ResCNN, TCN, OmniScaleCNN, eXplainable Convolutional neural network (XCM), Time-Series Transformer, LSTM-FCN, D-linear) with the original time series data and SOTA image analysis models (Pre-activation ResNet, DenseNet, DLA, Dual path networks (DPN), GoogleNet, Vision Transformer) with resized time series data. Finally, the concat-model and the paral-model architectures were designed for making predictions with both tabular data and time series data. It was determined that the concat-Model with TCN and ResCNN as the backbone exhibited the highest accuracy. It’s MAE for predicting Endpoint [P]/[C]/Temp reaches 2.153 × 10−3%, 6.413 × 10−3%, and 5.780°C, respectively, with field test’s MAE at 2.394 × 10−3%, 6.231 × 10−3%, and 7.679°C. Detailed results of the importance analysis for tabular data and time series are provided.