Theory and Applications of Time Series Analysis: Selected Contributions from ITISE 2019

Preface
Contents
Advanced Statistical and Mathematical Methods for Time Series Analysis
Random Forest Variable Selection for Sparse Vector Autoregressive Models
	1 Introduction
	2 State of the Art
		2.1 Feature Selection in Vector Autoregressive Models
		2.2 Random Forest for Feature Filtering
	3 Methodology and Data
		3.1 Methods
		3.2 Data: Urban Traffic Forecasting
	4 Results
	5 Discussion
	6 Conclusions
	References
Covariance Functions for Gaussian Laplacian Fields in Higher Dimension
	1 Introduction
	2 Frequency Domain Treatment of Stationary Fields in Higher Dimensions
		2.1 Covariance Functions of Laplacian Fields
		2.2 AR(p)
		2.3 ARMA Fields
	3 Appendix
	References
The Correspondence Between Stochastic Linear Difference and Differential Equations
	1 Introduction: The Discrete–Continuous Correspondence
	2 ARMA Estimation and the Effects of Over-Rapid Sampling
	3 Sinc Function Interpolation and Fourier Interpolation
	4 Discrete-Time and Continuous-Time Models
	5 ARMA Model and Its Continuous-Time CARMA Counterpart
	6 Stochastic Differential Equations Driven by Wiener Processes
	7 Summary and Conclusions
	References
New Test for a Random Walk Detection Based on the Arcsine Law
	1 Introduction
		1.1 Random Walk
		1.2 Ordinary Random Walk Test
		1.3 Random Walk Test for an AR(1) Process
	2 Efficiency Evaluation of the Proposed Test
		2.1 Gaussian Random Walk
		2.2 Gaussian Mixture Model
	3 Power Evaluation of the Proposed Test
		3.1 An AR(1) Process with the Gaussian Innovations
		3.2 An AR(1) Process with the Student-T Innovations
	4 Conclusions
	References
Econometric Models and Forecasting
On the Automatic Identification of Unobserved Components Models
	1 Introduction
	2 Unobserved Components Models
		2.1 Trend Components
		2.2 Seasonal Components
		2.3 Irregular Components
	3 State-Space Systems
	4 Automatic Forecasting Algorithm for UC
	5 Case Studies
		5.1 Monthly Average Temperatures in Madrid at El Retiro Weather Station
		5.2 Spanish Gross Domestic Product (GDP)
		5.3 Demand Database
	6 Conclusions
	References
Spatial Integration of Pig Meat Markets in the EU: Complex Network Analysis of Non-linear Price Relationships
	1 Introduction
	2 Data and Methods
		2.1 Data
		2.2 Filtering
		2.3 Non-linear Granger Causality Networks
		2.4 Network Measures
		2.5 Temporal Network Evolution
	3 Finite Sample Properties of the GAM-Test
	4 Empirical Analysis and Results
		4.1 Network Measures of Individual Node Connectivity
		4.2 Measures of Global Network Cohesiveness and Their Evolution
	5 Conclusion
	References
Comparative Study of Models for Forecasting Nigerian Stock Exchange Market Capitalization
	1 Introduction
	2 Literature Review
	3 Methodology/Material
		3.1 ARIMA Process
		3.2 The ARDL Process
		3.3 Test of Adequacy of Fitted Model
		3.4 Performance Evaluation
	4 Results
		4.1 Exploratory Data Analysis
		4.2 Unit Roots Test
		4.3 The ARIMA Process Result
		4.4 The ARDL Model Result
		4.5 Performance Evaluations of the Fitted Models
	5 Conclusion
	References
Industry Specifics of Models Predicting Financial Distress
	1 Introduction
	2 Literature Review in the Area of the Prediction Models
	3 Research Idea and Data
		3.1 Paper’s Idea and Used Methods
		3.2 Data Sample
	4 Results
	5 Conclusion
	References
Stochastic Volatility Models Predictive Relevance for Equity Markets
	1 Introduction
	2 Theory and Methodology
		2.1 Stochastic Volatility Models
		2.2 The Unobserved State Vector Using the Nonlinear Kalman Filter
	3 Stylized Facts of Volatility
		3.1 Tail Probabilities, the Power Law and Extreme Values
		3.2 Volatility Clustering
		3.3 Volatility Exhibits Persistence
		3.4 Volatility Is Mean Reverting
		3.5 Volatility Asymmetry (Leverage)
		3.6 Long Memory in Volatility
	4 European Examples: FTSE100 Index and Equinor Asset
		4.1 Equity Summaries
		4.2 The Stochastic Volatility Models for the European Equities
		4.3 Volatility Characteristics for the European Equities
		4.4 Step Ahead Volatility Predictions for European Equities
	5 Summary and Conclusions
	References
Empirical Test of the Balassa–Samuelson Effect in Selected African Countries
	1 Introduction
	2 Literature Review
		2.1 Introduction
		2.2 The Balassa–Samuelson Model
		2.3 Empirical Literature
	3 Methodology
		3.1 Model Specification
		3.2 Data Description
		3.3 Estimation Technique
	4 Estimation Results
		4.1 Panel Unit Root (Stationarity) Tests
		4.2 Cointegration Test Results
		4.3 Long-Run Coefficient
		4.4 Real Exchange Rate Misalignment
	5 Conclusion
	References
Energy Time Series Forecasting
End of Charge Detection by Processing Impedance Spectra of Batteries
	1 Introduction
	2 Data Generation
		2.1 Equipment
		2.2 Software
	3 Data Processing
		3.1 Processing of Raw Data
		3.2 Processing of Spectra Data
	4 Evaluation
	5 Conclusion and Outlook
	References
The Effect of Daylight Saving Time on Spanish Electrical Consumption
	1 Introduction
	2 DST Effects on Consumption: Simulation-Based Analysis
		2.1 Procedure
		2.2 Case Study
	3 Randomized Block Design and Paired Data Analysis
		3.1 Period of Study
		3.2 Exogenous Factors Removal
		3.3 Data Used in This Study
		3.4 Implemented Models
		3.5 Case Study
	4 Conclusions
	References
Wind Speed Forecasting Using Kernel Ridge Regression with Different Time Horizons
	1 Introduction
	2 Forecasting Models
		2.1 Persistent Model
		2.2 Least Squares Model
		2.3 Kernel Ridge Regression
	3 Methodology
	4 Results and Discussion
		4.1 One-Hour Ahead Time Horizon
		4.2 Twelve-Hours Ahead Time Horizon
		4.3 Day-Ahead (Twenty-Four Hours Ahead) Time Horizon
	5 Conclusion
	References
Applying a 1D-CNN Network to Electricity Load Forecasting
	1 Introduction
	2 Smart Meter Data
	3 Importance of Time Series Forecasting
	4 Convolutional Neural Networks
	5 Forecasting with CNNs
	6 Evalution of Different Network Structures and Training Parameters
		6.1 Development and Analysis of a Basic Forecaster
		6.2 Improvements to the Basic Forecaster
	7 Conclusion
	References
Long- and Short-Term Approaches for Power Consumption Prediction Using Neural Networks
	1 Introduction and Problem Description
	2 Data Description
		2.1 Power Consumption Dataset
		2.2 External Data
	3 Introduction to Neural Networks
		3.1 Long Short-Term Memory Neural Networks
		3.2 Convolutional Neural Networks
	4 Methodology
		4.1 Proposed Short-Term LSTM Network
		4.2 Proposed Convolutional Neural Network
		4.3 Improvements over the LSTM Network for Long-Term Time Series Forecasting
	5 Results
		5.1 Short-Term Time Series Forecasting
		5.2 Long-Term Time Series Forecasting
	6 Conclusions
	References
Forecasting Complex/Big Data Problems
Freedman\'s Paradox: A Solution Based on Normalized Entropy
	1 Introduction
	2 Maximum Entropy Estimators and Normalized Entropy
	3 Simulation Studies and Discussion
	4 Conclusions and Future Research
	References
Mining News Data for the Measurement and Prediction of Inflation Expectations
	1 Introduction
	2 Methodology
		2.1 Data
		2.2 Text Pre-processing
		2.3 Topic Modelling
	3 Results
	4 Application in Forecasting
	5 Conclusions
	Appendix 1: The List of Topics with Their Most Frequent Words
	References
Big Data: Forecasting and Control for Tourism Demand
	1 Introduction
	2 Literature Review
		2.1 Forecasting Methods Using Google Search Engines (Google Trends)
	3 Methodology
		3.1 Modelling and Forecasting Evaluation
	4 Data
	5 Empirical Results
	6 Conclusions
	References
Traffic Networks via Neural Networks: Description and Evolution
	1 Traffic Networks and Their Diverse Impact
	2 Neural Networks for Time Series Analysis
		2.1 The Three Neural Networks Chosen
		2.2 Network Design Specifics
	3 Data Filtering, Training and Simulations
		3.1 Images and Processing
		3.2 Filtering Cascade
	4 Training and Simulations
		4.1 Traffic Density Forecasting
	5 Traffic Signal Timings and Applications
		5.1 Stochastic Markov Model
		5.2 Stochastic Model to Simulate Traffic Dynamics
		5.3 Traffic Signal Assignment and Heat Map
	References
Time Series Analysis with Computational Intelligence
A Comparative Study on Machine Learning Techniques for Intense Convective Rainfall Events Forecasting
	1 Introduction
	2 Materials and Methods
		2.1 Dataset Description
		2.2 Machine Learning Techniques
	3 Results
	4 Conclusion
	References
Long Short-Term Memory Networks for the Prediction of Transformer Temperature for Energy Distribution Smart Grids
	1 Introduction
	2 Data and Methodology
		2.1 Data Acquisition
		2.2 Wiener-Granger Causality Analysis
		2.3 Recursive Neural Networks
	3 Results and Discussion
		3.1 Evaluation
		3.2 Candidate Variables
		3.3 Prediction Results
	4 Conclusions
	References
Deep Multilayer Perceptron for Knowledge Extraction: Understanding the Gardon de Mialet Flash Floods Modeling
	1 Introduction
	2 Materials and Methods
		2.1 Study Area: Location and General Description
		2.2 Database
		2.3 Artificial Neural Networks
		2.4 Extracting Information: KnoX Method
	3 Results
		3.1 Choice of Variables
		3.2 Model Selection
		3.3 Discharge Estimation
		3.4 Contributions of Input Variables
		3.5 Results: Contributions as a Function of Time Windows
		3.6 Results: Effects of the Bias
	4 Discussion
		4.1 Selecting a Model Type for Physical Knowledge Extraction
		4.2 Response Time and Contributions
		4.3 Bias Input Importance
	5 Conclusions and Perspectives
	References
Forecasting Short-Term and Medium-Term Time Series: A Comparison of Artificial Neural Networks and Fuzzy Models
	1 Introduction
	2 Related Works
	3 Concepts of LSTM and GRU
		3.1 Long Short-Term Memory
		3.2 Gated Recurrent Unit
	4 Fuzzy Time Series Models
		4.1 Fuzzy TS Model, Based on Fuzzified TS Values
		4.2 Fuzzy TS Model, Based on Fuzzified First Differences of TS Values
		4.3 Fuzzy TS Model, Based on Elementary Fuzzy Tendencies
	5 Experiments and Results
	6 Conclusion
	References
Inflation Rate Forecasting: Extreme Learning Machine as a Model Combination Method
	1 Introduction
	2 Extreme Learning Machine Method
	3 Proposed Benchmarks and Forecasting Strategy
		3.1 Benchmarks
		3.2 Forecasting Strategy
	4 Findings
		4.1 Pseudo-Real-Time Experiment
	5 Concluding Remarks
	References
Time Series Analysis and Prediction in Other Real Problems
Load Forecast by Multi-Task Learning Models: Designed for a New Collaborative World
	1 Introduction
	2 Objective
	3 Multi-Task Learning Approach
	4 Architecture Differences
	5 The Classical Hilbert Space Approach
		5.1 Projection Theorem
		5.2 Parallel Processing Implementation
	6 Case Study
		6.1 The Challenge
		6.2 The Proposed Solution
	7 Conclusions
	References
Power Transformer Forecasting in Smart Grids Using NARX Neural Networks
	1 Introduction
	2 System Identification Modeling
	3 NARX Neural Network for Power Transformer Temperature Prediction
	4 Power Transformation Center Datasets
	5 Electrical Measurement Analysis from Power Transformers
		5.1 Correlation Analysis
		5.2 Granger Causality-Based Analysis
	6 NARX Implementation and Evaluation Experiments
	7 Conclusion
	References
Short-Term Forecast of Emergency Departments Visits Through Calendar Selection
	1 Introduction
		1.1 Crowding and Boarding
		1.2 Forecast Motivation
	2 Methodology
		2.1 Calendar Conditions
	3 Model Evaluation
		3.1 Results
		3.2 Comparison to Other Models
	4 Conclusion
	References
Discordant Observation Modelling
	1 Introduction
	2 Related Work
	3 Volatility Modelling
	4 Evaluation
		4.1 Text Analysis
		4.2 Volatility
		4.3 Normality
		4.4 Unit Root Test
		4.5 Trend and Seasonality
		4.6 Goodness of Fit
		4.7 Models
	5 Conclusion
	References
Applying Diebold–Mariano Test for Performance Evaluation Between Individual and Hybrid Time-Series Models for Modeling Bivariate Time-Series Data and Forecasting the Unemployment Rate in the USA
	1 Introduction and Motivation
	2 Materials and Methods
		2.1 The Hybrid ARMAX-GARCH-GED Forecasting Model
		2.2 The Zhang Hybrid Methodology
		2.3 Forecasts Evaluation
	3 Case Study (Data Sets in the Experiment)
		3.1 Fitting the Hybrid ARMAX-GARCHX-GED Model
	4 Evaluation of Forecasting Performance
		4.1 Loss Function Criteria
		4.2 Forecasting Evaluation Based on DM Test
	5 Conclusions
	References
Author Index