Theory and Applications of Time Series Analysis and Forecasting : Selected Contributions From ITISE 2021

This book presents a selection of peer-reviewed contributions on the latest developments in time series analysis and forecasting, presented at the 7th International Conference on Time Series and Forecasting, ITISE 2021, held in Gran Canaria, Spain, July 19-21, 2021. It is divided into four parts. The first part addresses general modern methods and theoretical aspects of time series analysis and forecasting, while the remaining three parts focus on forecasting methods in econometrics, time series forecasting and prediction, and numerous other real-world applications. Covering a broad range of topics, the book will give readers a modern perspective on the subject. The ITISE conference series provides a forum for scientists, engineers, educators and students to discuss the latest advances and implementations in the foundations, theory, models and applications of time series analysis and forecasting. It focuses on interdisciplinary research encompassing computer science, mathematics, statistics and econometrics. Preface Contents Part I Theoretical Aspects of Time Series An Improved Forecasting and Detection of Structural Breaks in Time Series Using Fuzzy Techniques 1 Introduction 2 Processing Time Series Using Fuzzy Modeling Methods 2.1 Time Series Decomposition 2.2 Fuzzy Transform (F-Transform) 2.3 Fuzzy Natural Logic 3 Forecasting Time Series 4 Detection of Structural Breaks in Time Series 5 Demonstration of Nonstatistical Forecast and Detection of Structural Breaks on Real Data 5.1 ARIMA Model 5.2 Forecasting Using LFL Forecaster 5.3 Demonstration of Found Structural Breaks 6 Conclusion References Anomaly Detection Algorithm Using a Hybrid Modelling Approach for Energy Consumption Time Series 1 Introduction 2 Our Definitions 3 Our Hybrid Model 4 Results 5 Summary References Unit Root Test Combination via Random Forests 1 Introduction 2 Unit Root Tests 2.1 Non-seasonal Unit Roots 2.2 Seasonal Unit Roots 3 Random Forests 3.1 Classical Random Forests 3.2 Conditional Random Forests 4 Test Evaluation 5 Results 6 Summary References Probabilistic Forecasting of Seasonal Time Series 1 Introduction 2 Probabilistic Seasonal Time Series Forecasting 2.1 Seasonal Time Series 2.2 Seasonal Probabilistic Forecasting 3 The P-F2C Forecaster 3.1 Co-clustering of Time Series: A Probabilistic Model 3.2 Predict the Next Type of Seasons 3.3 Select the Best Parameters (Portfolio) 4 Illustration on a Synthetic Dataset 4.1 The Data Generated 4.2 Grid Probabilistic Forecasts 5 Experiments 5.1 Experimental Protocol 5.2 Parameters' Sensitivity 5.3 P-F2C and PP-F2C vs Opponents 6 Conclusion Annexes References Nonstatistical Methods for Analysis, Forecasting, and Mining Time Series 1 Introduction 2 Fuzzy Transform 3 Fuzzy Natural Logic 4 Analysis of Time Series 5 Forecasting Time Series 6 Mining Information from Time Series 7 Conclusion References PMF Forecasting for Count Processes: A Comprehensive Performance Analysis 1 Introduction 2 Coherent and Approximate PMF Forecasting 3 Performance Evaluation: A Critical Literature Review 4 Results from a Comprehensive Simulation Study 4.1 General Results 4.2 Performance of Coherent Forecasting 4.3 Performance of Approximate Forecasting 5 Application: PMF Forecasting of Transaction Counts 6 Conclusions References A Novel First-Order Autoregressive Moving Average Model to Analyze Discrete-Time Series Irregularly Observed 1 Introduction 2 Model Formulation 3 An Irregular Observed First-Order Autoregressive Moving Average Model 3.1 Properties 3.2 State-Space Representation 3.3 Prediction 4 Maximum Likelihood Estimation 5 Monte Carlo Experiments 5.1 Performance Measures 5.2 Simulation Results 6 Applications 6.1 Lung Function of an Asthma Patient 6.2 Light Curve of an Astronomical Object 7 Conclusions References Part II Econometric and Forecasting Using Natural Language Processing to Measure COVID-19-Induced Economic Policy Uncertainty for Canada and the USA 1 Introduction 2 The Development of the Baker-Bloom-Davis EPU (BBD-EPU) 3 Constructing the EPU-NLP Index: Data, Methodology, and Algorithms 3.1 The RAKE (Rapid Automatic Keyword Extraction) Algorithm 3.2 The BERT, RoBERTa, and SBERT Algorithms 3.3 GrapeNLP Grammar 3.4 Calculating the EPU-NLP 4 Testing the Model 5 Conclusion Appendix References Asymptotic Expansions for Market Risk Assessment: Evidence in Energy and Commodity Indices 1 Introduction 2 Methodology 2.1 Gram-Charlier Expansion 2.2 Student's t Expansion 2.3 Model and Maximum Likelihood Estimation 2.4 Risk Measures 2.5 Backtesting 3 Empirical Results 3.1 Data 3.2 In-Sample Analysis 3.3 Backtesting 4 Conclusions References Predicting Housing Prices for Spanish Regions 1 Introduction and Motivation 2 Previous Experiences and Evidence: Literature Review of Housing Price Prediction 3 Theoretical Basis 4 Data 5 Methodology 5.1 Empirical Strategy to Estimate the Model 5.2 Forecast Methodology 6 Results Empirical Evidence 6.1 Discussion 7 Conclusions References Optimal Combination Forecast for Bitcoin Dollars Time Series 1 Introduction 2 Method 2.1 Exponential Smoothing Model 2.2 Optimization 2.3 ARIMA Model 2.4 Artificial Neural Network 2.5 Forecast's Combination 2.6 Measuring Forecast Accuracy 3 Results and Discussion 3.1 Exponential Smoothing Model Result 3.2 ARIMA Model Result 3.3 Artificial Neural Networks 3.4 Combination Model Result 4 Conclusion References The Impact of the Hungarian Retail Debt Program 1 Introduction 2 The Hungarian Retail Debt Program 2.1 Main Objectives 2.2 The Retail Debt Portfolio 3 The Historical Cost of Retail Debt 3.1 Methodology 3.2 Results 4 Forecasting the Important Macroeconomic Variables 4.1 Methodology 4.2 Results 5 The Future of the Retail Debt Program 5.1 Estimation of the Factors Driving the Outstanding Amount of Retail Debt 5.2 Simulation and Results 6 Conclusion References Predicting the Exchange Rate Path: The Importance of Using Up-to-Date Observations in the Forecasts 1 Introduction 2 Theory 3 Results 4 Conclusions References Part III Time Series Prediction Applications Development of Algorithm for Forecasting System Software The List of Acronyms 1 Introduction 2 Data and Materials 3 The Review of Ensembling Time Series and Neural Network System (ET-System) for Forecasting Covid-19 Cases and Waves for Infection Cases 3.1 The Algorithm Schema of the Ensembling Time Series and Neural Network System (ET-System) 3.2 The Scheme of the Algorithm for Dynamic Prioritizer 3.3 The Scheme for Bagging and Bootstrapping the NNAR Model for Improving Forecasting of the Waves of Infection Cases 3.4 Design of the Software for Forecasting 4 Results 5 Conclusions and Further Research Appendix References Forecasting High-Frequency Electricity Demand in Uruguay 1 Introduction 2 Methodological Approach 2.1 General Model and the Treatment of Special Days 2.2 A Non-linear Approach to Model the Effect of Climate Variables 3 The Data 4 Results 4.1 Modelization of Special Days 4.2 Nonlinear Modelization of the Effect of Climate Variables 4.3 Predictive Evaluation 4.4 Evaluation of the Prediction System During the Health Emergency 5 Main Conclusions References Day-Ahead Electricity Load Prediction Based on Calendar Features and Temporal Convolutional Networks Acronyms 1 Introduction 2 Data 2.1 Electricity Load 2.2 Electricity Load Prediction 2.3 Calendar Data 3 Data Analysis 4 Model Architecture 4.1 Feedforward Network Based on Calendar Features 4.2 Temporal Convolutional Network 4.3 Hybrid Model 5 Training and Evaluation Set 6 Results 7 Conclusion References Network Security Situation Awareness Forecasting Based on Neural Networks 1 Introduction 2 Related Works 3 Methodology 3.1 Dataset 3.2 Method Description 3.3 Neural Networks 3.4 Statistical Methods 4 Experiment Evaluation 5 Results and Discussion 6 Conclusion and Future Works References Part IV Advanced Applications in Time Series Analysis Modeling Covid-19 Contagion Dynamics: Time-Series Analysis Across Different Countries and Subperiods 1 Introduction 2 The Markov-Switching Time-Series Models 3 Empirical Results for Early Dynamics 3.1 Italy (Sample: February 22–May 31, 2020) 3.2 Germany (Sample: January 28–May 31, 2020) 3.3 The United Kingdom (Sample: February 24–May 31, 2020) 3.4 Russia (March 3–May 31, 2020) 3.5 Results Summary 4 Empirical Results for Italy: The Second and the Third Peak 4.1 The Second Peak (Sample: October 1, 2020–January 31, 2021) 4.2 The Third Peak (Sample: February 1–May 15, 2021) 4.3 Comparison of Two Subsamples 5 Concluding Remarks References Diffusion of Renewable Energy for Electricity: An Analysis for Leading Countries 1 Introduction 2 Motivation: Energy Trends 3 Background 4 Model 4.1 Estimation and Model Selection 5 Application 6 Discussion References The State and Perspectives of Employment in the Water Transport System of the Republic of Croatia 1 Introduction 2 Theoretical Framework 3 Descriptive Analysis of Employment in Water Transport in the EU 3.1 Descriptive Analysis of Employment in Sea Transport in the EU 3.2 Descriptive Analysis of Employment in the EU Inland Waterway Transport 4 Data and Research Methodology 5 Research Result and Discussion 6 Conclusion References Reversed STIRPAT Modeling: The Role of CO2 Emissions, Population, and Technology for a Growing Affluence 1 Introduction 2 Perspectives of Causality in the STIRPAT Model Approach 3 Methodological Remarks 4 Empirical Application 4.1 Granger-Causality, Non-stationarity, and Cointegration 4.2 Reversed STIRPAT 5 Discussion of Results 6 Concluding Remarks Appendix References