Skip to main content
Top

2023 | Book

Big Data Analytics for Smart Transport and Healthcare Systems

Authors: Saeid Pourroostaei Ardakani, Ali Cheshmehzangi

Publisher: Springer Nature Singapore

Book Series : Urban Sustainability

insite
SEARCH

About this book

This book aims to introduce big data solutions in urban sustainability applications—mainly smart transportation and healthcare systems. It focuses on machine learning techniques and data processing approaches which have the capacity to handle/process huge, live, and complex datasets in real-time transportation and healthcare applications. For this, several state-of-the-art data processing approaches including data pre-processing, classification, regression, and clustering are introduced, tested, and evaluated to highlight their benefits and constraints where data is sensitive, real-time, and/or semi-structured.

Table of Contents

Frontmatter
Chapter 1. The Role of Big Data Analytics in Urban Systems: Review and Prospect for Smart Transport and Healthcare Systems
Abstract
This introduction chapter provides an overview of the idea, aim, and objectives of the book. It delves into the importance of Big Data as a tool in the current information age. The chapter starts with an overview of Big Data analytics for urban systems and follows the discussions from the sector-based perspectives. It then explores Big Data Applications (BDA) in two key areas of smart transportation and healthcare, particularly in cities and as part of (smart) urban systems. After providing a review of prospects about Big Data analytics in these two sectors, the chapter introduces the book structure and all case study chapters. This chapter shares an overall picture of the book to readers before we delve into global case study examples.
Saeid Pourroostaei Ardakani, Ali Cheshmehzangi

Smart Transport

Frontmatter
Chapter 2. Big Data Analysis for an Optimised Classification for Flight : Prediction Analysis Using Machine  Classifiers
Abstract
Accurate flight delay forecasting is critical for establishing a more efficient airline industry. A smart system in place would help reducing the negative impacts of such delays. Machine learning-enabled Big data solutions have been widely utilized in recent studies to anticipate aircraft delays. They need a data pre-processing to understand and grasp the relevance of each data attribute. The results of data attribute relevance are used to filter out data deemed relevant to aircraft delays and eliminate the data that was redundant and unsuitable for analysis. This chapter trains linear and polynomial regression models to predict the delay time of a flight. The data analysis algorithm runs on a well-known dataset, which comprises flight data from more than ten US airlines from 2009 to 2019. The result indicates that 97.04\(\%\) of the predicted result has a difference of fewer than 15 min between the actual value.
Saeid Pourroostaei Ardakani, Ali Cheshmehzangi
Chapter 3. On-Board Unit Freight Transport Data Analysis and Prediction: Big Data Analysis for Data Pre-processing and Result
Abstract
Traffic prediction is a complex urban problem that needs to collect and analyse big data and train accurate machine learning models. Solving it will contribute to society’s efficient production and development, particularly in optimising urban systems. This chapter aims to build a precise model to predict the number of freight vehicles in future timestamps based on historical data provided On-Board Unit (OBU) datasets in Belgium’s road networks. We contributed to two novel solutions to solve the time series prediction task. The first contribution is preprocessing the data using SparkSQL and generating nine features from the prediction timestamps. The second is LSTM and LSTM + FCN deep learning models with tuned parameters and training with the newly generated features. The results of our LSTM model are more accurate than those of other models on this dataset that we have already examined, reaching an accuracy of 99.89%. The preprocessing stage is proven to be vital for the performance of the LSTM model. As a result, models using our preprocessed nine features are much more accurate on this freight transportation prediction task. Discussions are also raised to understand better freight transport prediction tasks from the data and results points of view.
Saeid Pourroostaei Ardakani, Ali Cheshmehzangi
Chapter 4. Data-Driven Multi-target Prediction Analysis for Driving Pattern Recognition: A Machine Learning Approach to Enhance Prediction Accuracy
Abstract
The driving pattern is critical in life quality enhancement, road traffic minimisation, and transportation risk reduction. It includes several parameters, such as the car’s speed, traffic, weather, and road status. This article investigates the correlations between driving attributes and proposes a multi-target prediction model to recognise driving patterns. For this, a pre-processing data approach, including Pearson’s Correlation Coefficient, Quantile Discretisation, and Hybrid Data Resampling, is used to reduce data feature dimensions and remove meaningless variables. The Random Forest technique predicts four targets, including vehicle speed, rain intensity, driver’s well-being, and driver’s rush. At the same time, the K-Nearest Neighbours algorithm (KNN) groups the prediction results and forms the driving patterns. The performance of the proposed multi-target prediction model is compared with four classifiers, including Multilayer Perceptron, Decision Tree, Multinomial Logistics Regression, and LR one versus rest. According to the results, the Random Forest model outperforms the benchmarks regarding prediction accuracy. The findings of this chapter help optimise prediction accuracy that could then be used for urban transportation system optimisation.
Saeid Pourroostaei Ardakani, Ali Cheshmehzangi
Chapter 5. A Predictive Data Analysis for Traffic Accidents: Real-Time Data Use for Mobility Improvement and Accident Reduction
Abstract
Road traffic accidents are a significant component of human and economic losses. In order to implement effective policies or carry out safe traffic travel and improve the road system’s safety, it is essential to identify the data patterns in the dataset for extracting the main influencing features associated with traffic accidents. The analysis of traffic accidents is complex as they affect each other and are also affected by many other factors. We used the US Accidents dataset, covering 2.8 million accident records in 49 states in the United States from February 2016 to December 2021. Maximum Relevance (MR) is used to get feature relevance by obtaining Mutual Information (MI) of features. The traffic accident predictive models are expected to be more accurate by using feature relevance to assist people in making real-time transportation decisions to improve mobility and reduce accidents. The findings help improve urban transportation network and systems at multiple spatial levels.
Saeid Pourroostaei Ardakani, Ali Cheshmehzangi

Smart Healthcare

Frontmatter
Chapter 6. Healthcare Infrastructure Development and Pandemic Prevention: An Optimal Model for Healthcare Investment Using Big Data
Abstract
This chapter proposes an optimal model for predicting the minimum healthcare investment required for pandemic prevention. It focuses on the COVID-19 pandemic as a recent example to further analyse the role of healthcare infrastructure development. It highlights the relationship between the control of COVID-19 and infrastructure development using linear regression, SVR, KNN, and decision tree models. By analysing the impact of each feature on the results, we select appropriate attributes and data processing methods. The findings of this study demonstrate that KNN is the best model with the highest training score of 0.655, while the training result of the decision tree is the worst, and the score is no more than 0.5. Lastly, the study highlights how big data could be used to improve the availability and development of urban critical infrastructures, such as healthcare infrastructure. An optimal model is suggested as part of the conclusion of this study.
Saeid Pourroostaei Ardakani, Ali Cheshmehzangi
Chapter 7. Big Data for Social Media Analysis During the COVID-19 Pandemic: An Emotion  Based on Influences from Social Networks
Abstract
Since the beginning of the COVID-19 outbreak, the situation has soon become the most popular topic across social networks worldwide. For almost three years, news and social media were covered with information, updates, and daily/regular reports on the COVID-19 pandemic. When face-to-face activities were restricted, people started to use social networks more than before. The situation led to the empowerment of digital media, such as social media and networks. These platforms became the leading online social hubs for people to express their feelings and record their daily lives during the pandemic. In this study, we analyzed this critical societal change as a major topic, mainly related to the pandemic on Weibo, which is the most popular social media platform in China. The study is focused on one particular context allowing in-depth data analysis related to China’s societal impact due to the COVID-19 pandemic. The study also investigates the relationship between COVID-19 trends or topics and public sentiments on social networks. Machine learning is used to verify the correlation between emotion on social media and the COVID-19 pandemic trends. The study concludes with a prediction model using big data for public sentiments on social media.
Saeid Pourroostaei Ardakani, Ali Cheshmehzangi
Chapter 8. Big Data-Enabled Time Series Analysis for Climate Change Analysis in Brazil: An Artificial Neural Network Machine Learning Model
Abstract
Climate data is an essential kind of data for humans in the world. Improving the ability to forecast the climate will contribute to the development of many industries, such as agriculture and shipping. In this project, we use the climate data in Brazil from 2000 to 2020. Attributes of the data mainly are date and time, temperature, precipitation, wind speed, and the province in which these data are measured. This study aims to classify these climate data and analyze the changing climate trends in the same province. An artificial neural network is established as the model in this project to implement this objective. The performance shows that this model can complete this classification task.
Saeid Pourroostaei Ardakani, Ali Cheshmehzangi
Chapter 9. Optimized Clustering Model for Healthcare Sentiments on Twitter: A Big Data Analysis Approach
Abstract
Social media, such as Twitter, typically stores a large amount of user-generated content regarding different aspects of society. These contents include social events, e-commerce products, healthcare, etc. This chapter proposes a best-fitted clustering method to classify sentiment samples related to healthcare topics. Thus, we examine other clustering models with keyword extraction methods on the real healthcare datasets collected from Twitter. The experiment results indicate that self-organized map model with the TF-IDF extraction method can achieve the best clustering accuracy. Moreover, the optimized model can have great potential to handle large-scale data in real practice.
Saeid Pourroostaei Ardakani, Ali Cheshmehzangi
Chapter 10. Big Data Analytics and the Future of Smart Transport and Healthcare Systems
Abstract
Urban experimentation has long become a platform of city transitions, where we see opportunities for smart development. This applies to many sectors, including smart transportation and healthcare. In this concluding chapter, we delve into some key sectoral contributions before highlighting some of the lessons learnt. The power of Big Data analytics, through both correlational and predictive analysis, indicates various ways of facilitating knowledge for institutional development, governance enhancement, and operational optimisation. In both smart transport and smart healthcare systems, we see these transitions happening at a gradual pace. Thus, we believe these two sectors are leading the Big Data analytics research and practice, particularly in the context of smartness and smart city development. The chapter also summarises some of the key lessons from all case study chapters.
Saeid Pourroostaei Ardakani, Ali Cheshmehzangi
Backmatter
Metadata
Title
Big Data Analytics for Smart Transport and Healthcare Systems
Authors
Saeid Pourroostaei Ardakani
Ali Cheshmehzangi
Copyright Year
2023
Publisher
Springer Nature Singapore
Electronic ISBN
978-981-9966-20-2
Print ISBN
978-981-9966-19-6
DOI
https://doi.org/10.1007/978-981-99-6620-2

Premium Partner