Contents


Preface

A time series is a sequence of data points that are measured and ordered over uniform time intervals. Examples include: household energy consumption, heart rates, hourly air temperature measurements, daily company stock prices, monthly auto sales, and yearly sales figures. Time series analysis is a statistical technique that analyzes time series and then extracts meaningful statistics, characteristics, and insights from the data. Some applications include identifying cyclical changes in sales of a product, forecasting business budget based on historical trends, and studying employee turnover data and employee training data to determine if there is any dependence of employee training programs on employee turnover rates over time. Thus, analyzing time series data is an important topic that has gained significant interest in the past decade. However, due to the data pattern complexity in different domains and industries nowadays, time series analysis has become very challenging and this has inspired domain scientists and practitioners to develop and apply the updated methodologies and applications to solve real-world time series

problems such as forecasting, classification, and feature extraction.

Section 1: Mining Complex Patterns in Time Series Data

the content easily.

To address the above issues, the book editor has carefully selected a number of domain experts' papers that focus on data, methods, and applications on time series analysis that provide our readers with the latest information, developments, and trends in this research area. Specifically, this book is divided into three sections: (1) Mining complex patterns in time series data, (2) Deep neural networks for time series analytics, and (3) Time Series forecasting in real-world problems. Each section includes two chapters. Each chapter starts with motivational background, technical foundations, specific methodologies, and then ends with examples and case studies to explain the concepts and techniques that help our readers understand

• Process Fault Diagnosis for Continuous Dynamic Systems over Multivariate Time Series. Use grey level co-occurrence matrices and local binary patterns to

• Fuzzy Forecast Based on Fuzzy Time Series. Use fuzzy time series for interval

• Training Deep Neural Networks with Reinforcement Learning for Time Series Forecasting. Combine multiple restricted Boltzmann machines (RBMs) and multilayer perception (MLP) and use stochastic gradient ascent (SGA) training

• CNN Approaches for Time Series Classification. Use Convolutional Neural

extract features from time series for process fault diagnosis.

prediction and long-term significance level analysis.

Network (CNN) models for time series classification.

Section 2: Deep Neural Networks for Time Series Analytics

algorithm for time series forecasting.
