Research Projects for MSc AI and Data Science

Current cohort: Sep 2023

Welcome to the MSc research project on urban big data analytics, I'm delighted to have you on board as we delve into intriguing topics in urban science and crime science and I believe you will have confidence in your proficiency across most aspects of urban big data analytics and meet the requirements for the research reports upon completing this learning journey. While uni emphasizes the importance of avoiding academic misconduct each academic year, I must also reemphasise that any form of misconduct is unacceptable for this project. Please ensure thorough preparation and patience as you focus on each stage, which will lead to the completion of the report with high quality, ultimately contributing to the attainment of your MSc degree.

---

1 Research proposal

A research proposal outlines a roadmap for the research project, including the objectives, methodology, timeline, and expected outcomes. In urban analytics, a research proposal aims to delve into urban issues by employing data science and AI methodologies to extract valuable insights pertaining to the multifaceted nature of urban environments.

Here are the typical sections included in a research proposal:

• Title: A concise title that captures the essence of the research project.

• Introduction: This section should include Background information, Research gaps, Research questions or hypothesis, and Objectives or aims for the research projects.
  • Background information: It provides context for the study, including relevant literature, previous research, and the current state of knowledge in the field. It helps others understand why the research is important.
  • Research gaps: It refers to the research area or question where existing research has not adequately addressed or answered. Research gaps can arise for various reasons, such as limitations in previous studies (from literature), emerging trends or technologies and varying societal needs that have not been explored.
  • Research questions or hypothesis: It should clearly state the main research question or hypothesis that the study seeks to address.
  • Objectives or aims: It explains what the research aims to achieve or what specific concern you hope to solve and why these goals are important within the scope of the proposed study.
• Literature Review: A review of relevant literature related to the research topic. This section demonstrates your understanding of existing concepts, findings, and methodologies, particularly the summary of current research challenges and gaps generated from the literature review.
• Methodology: Detailed explanation of the research methodology, including data sources, and methods (e.g., data preprocessing techniques, variable measurement and the selected models).
• Expected Outcomes: Anticipated outcome results of the research and how they will contribute to the existing body of knowledge in urban analytics.
• Timeline: A timeline outlining the different stages of the research project, including data collection, analysis, writing, and submission.
• References: A list of references cited in the proposal, following a specific citation style (e.g., APA, MLA).

Specifically, Table 1 lists all the important research elements for research proposals in urban analytics. Please incorporate all the elements in your research proposal and clearly deliver this table as Appdedix in the proposal.

Table 1 The checklist of the important elements in research proposal.

Element	Description	Example
Data type	Data types refer to the formats of information collected. In urban science, data can be categorised into various types for different intentions. For example, Crime data can be collected by crime survey data, policing recorded data, sel-report data and so on. Urban mobility data can be categorised into mobile phone call detail records, mobile phone GPS data, underground smart card data, WiFi data, social media data and so on.	Underground smart card data, policing recorded data
Data resource	Data resources are the various sources you collected the data under different usage licences (e.g., education and research licenses). Please read the data use policy carefully if you get access to the open data.	London datastore
Independents/features/predictors(X)	Features or independent variables are the attributes of the data that are used to predict the outcome.	Population mobility variables (measured by travel behaviours from smart card data)
Dependents/targets/responses(y)	Target variable or dependent variable predict or understand based on the feature variables. For supervised learning tasks, y typically consists of the labels or responses (e.g., can be a column) associated with each set of feature variables in the dataset.	Theft counts
Spatial unit of analysis	The spatial unit of analysis refers to the geographic level or scale in the research analysis. It defines the spatial resolution or granularity in understanding spatial patterns. Clearly defining the geographical unit of analysis can help to avoid the Ecological Fallacy in the research findings.	Lower Super Output Area (LSOA)
Temporal unit of analysis	Temporal unit of analysis refers to the time scale or interval in analysis, e.g., the examination of temporal patterns, trends, or relationships. It can categorised into hourly, daily, weekly, monthly or yearly. In some prediction tasks, it emphasises the predicting power in the time scale of the trained model, e.g., the model can predict the next week (week-level) for each LSOA.	Monthly
Study area and city	It means the specific urban areas of some select cities as the case study in the analysis, e.g., City of London areas in Greater London.	All urban areas in Greater London
Observation period	It means the temporal period of the observation in the experimental analysis, e.g., the observation period covers 2021 to 2022 (two years).	2021 year
Model/method	The main method/model will be used or trained for solving the research questions, such as some statistical models or machine learning models.	Random Forest regressor

The starting date of the research project for this cohort is May 30 2024.

2 Recommended reading

The recommended reading section offers review papers and empirical works to aid in grasping the fundamental concepts, methodologies, and data types employed within specific topics. However, due to the proliferation of advanced methods (especially AI and Data Science) and emerging data types in these hot research topics, it may not cover all relevant literature in each particular research domain. It is strongly advised to further explore specific topics through Google Scholar or the university library for comprehensive understanding. To clarify, each number following the literature reference corresponds to the related topics index, which can be located in the appendix section.

• Mandalapu, Varun, Lavanya Elluri, Piyush Vyas, and Nirmalya Roy. “Crime prediction using machine learning and deep learning: A systematic review and future directions.” IEEE Access (2023). related topics: 1, 2, 3
• Kounadi, Ourania, Alina Ristea, Adelson Araujo, and Michael Leitner. “A systematic review on spatial crime forecasting.” Crime science 9 (2020): 1-22. related topics: 1, 2, 3
• Barbosa, Hugo, Marc Barthelemy, Gourab Ghoshal, Charlotte R. James, Maxime Lenormand, Thomas Louail, Ronaldo Menezes, José J. Ramasco, Filippo Simini, and Marcello Tomasini. “Human mobility: Models and applications.” Physics Reports 734 (2018): 1-74. related topics: 5, 6, 7, 8, 9, 10
• Pappalardo, Luca, Ed Manley, Vedran Sekara, and Laura Alessandretti. “Future directions in human mobility science.” Nature Computational Science 3, no. 7 (2023): 588-600. related topics: 5, 6, 7, 8, 9, 10
• Dong, Lei, Fabio Duarte, Gilles Duranton, Paolo Santi, Marc Barthelemy, Michael Batty, Luís Bettencourt et al. “Defining a city—delineating urban areas using cell-phone data.” Nature Cities 1, no. 2 (2024): 117-125. related topics: 5, 6, 7, 8, 9, 10
• Jiang, Yanxiao, Yuyang Zhang, Yu Liu, and Zhou Huang. “A review of urban vitality research in the Chinese world.” Transactions in Urban Data, Science, and Technology 2, no. 2-3 (2023): 81-99. related topics: 9, 10
• Okmi, Mohammed, Lip Yee Por, Tan Fong Ang, and Chin Soon Ku. “Mobile phone data: A survey of techniques, features, and applications.” Sensors 23, no. 2 (2023): 908. related topics: 5, 6, 7, 8, 9, 10
• Moraffah, Raha, Paras Sheth, Mansooreh Karami, Anchit Bhattacharya, Qianru Wang, Anique Tahir, Adrienne Raglin, and Huan Liu. “Causal inference for time series analysis: Problems, methods and evaluation.” Knowledge and Information Systems 63 (2021): 3041-3085. related topics: 4
• Yao, Liuyi, Zhixuan Chu, Sheng Li, Yaliang Li, Jing Gao, and Aidong Zhang. “A survey on causal inference.” ACM Transactions on Knowledge Discovery from Data (TKDD) 15, no. 5 (2021): 1-46. related topics: 5
• Hu, Tao, Siqin Wang, Bing She, Mengxi Zhang, Xiao Huang, Yunhe Cui, Jacob Khuri et al. “Human mobility data in the COVID-19 pandemic: characteristics, applications, and challenges.” International Journal of Digital Earth 14, no. 9 (2021): 1126-1147. related topics: 5
• Chang, Serina, Emma Pierson, Pang Wei Koh, Jaline Gerardin, Beth Redbird, David Grusky, and Jure Leskovec. “Mobility network models of COVID-19 explain inequities and inform reopening.” Nature 589, no. 7840 (2021): 82-87. related topics: 6
• Wang, Qi, Nolan Edward Phillips, Mario L. Small, and Robert J. Sampson. “Urban mobility and neighborhood isolation in America’s 50 largest cities.” Proceedings of the National Academy of Sciences 115, no. 30 (2018): 7735-7740. related topics: 6
• Ma, Xiaolei, Jiyu Zhang, Chuan Ding, and Yunpeng Wang. “A geographically and temporally weighted regression model to explore the spatiotemporal influence of built environment on transit ridership.” Computers, Environment and Urban Systems 70 (2018): 113-124. related topics: 7
• Alattar, Mohammad Anwar, Caitlin Cottrill, and Mark Beecroft. “Modelling cyclists’ route choice using Strava and OSMnx: A case study of the City of Glasgow.” Transportation research interdisciplinary perspectives 9 (2021): 100301 related topics: 7
• Gao, Qi-Li, Yang Yue, Chen Zhong, Jinzhou Cao, Wei Tu, and Qing-Quan Li. “Revealing transport inequality from an activity space perspective: A study based on human mobility data.” Cities 131 (2022): 104036. related topics: 9, 10
• Aldred, Rachel, Anna Goodman, John Gulliver, and James Woodcock. “Cycling injury risk in London: A case-control study exploring the impact of cycle volumes, motor vehicle volumes, and road characteristics including speed limits.” Accident Analysis & Prevention 117 (2018): 75-84. related topics: 8
• Li, Jintai, and Zhan Zhao. “Impact of COVID-19 travel-restriction policies on road traffic accident patterns with emphasis on cyclists: A case study of New York City.” Accident Analysis & Prevention 167 (2022): 106586. related topics: 8
• Cui, Yu, Chuishi Meng, Qing He, and Jing Gao. “Forecasting current and next trip purpose with social media data and Google places.” Transportation Research Part C: Emerging Technologies 97 (2018): 159-174. related topics: 5, 7
• Liu, Shaojun, Ling Zhang, Yi Long, Yao Long, and Mianhao Xu. “A new urban vitality analysis and evaluation framework based on human activity modeling using multi-source big data.” ISPRS International Journal of Geo-Information 9, no. 11 (2020): 617. related topics: 9
• Meng, Yuan, and Hanfa Xing. “Exploring the relationship between landscape characteristics and urban vibrancy: A case study using morphology and review data.” Cities 95 (2019): 102389. related topics: 9, 10
• Xu, Fengli, Yong Li, Depeng Jin, Jianhua Lu, and Chaoming Song. “Emergence of urban growth patterns from human mobility behavior.” Nature Computational Science 1, no. 12 (2021): 791-800. related topics: 5, 10
• Qin, Quan, Shishuo Xu, Mingyi Du, and Songnian Li. “Identifying urban functional zones by capturing multi-spatial distribution patterns of points of interest.” International Journal of Digital Earth 15, no. 1 (2022): 2468-2494. related topics: 10
• Choi, Junyong, Wonjun No, Minju Park, and Youngchul Kim. “Inferring land use from spatialtemporal taxi ride data.” Applied Geography 142 (2022): 102688. related topics: 10

3 Data sources

Crime data

Nowadays, crime data is readily available through numerous resources for various purposes. Several resources from different cities in the UK and US are listed below for your reference. Please take note of the data usage license and consider the data quality, particularly regarding spatial and temporal resolution issues, as detailed in the provided resources. Additionally, some city open data portals offer additional urban data that can be linked to the crime data for further analysis.

• England cities, UK
• San Francisco, CA, US
• Los Angeles, CA, US
• Phienix, AZ, US
• Minneapolis, MN, US
• Chicago, IL, US
• Baltimore, MD, US
• Boston, MA, US
• Vancouver, CA

Urban data

Urban data is accessible through various resources, encompassing socio-economic data, population statistics, transportation data, geographical boundaries, and other urban environmental data. Below are several urban data repositories. Please be aware that some data sources require registration and obtaining an educational license for usage. You are encouraged to delve deeper into additional data resources or utilise your own data sets.

• Office for National Statistics, UK
• London Datastore
• Digimap
• UK Data Service
• Transport for London Open Data
• OpenStreetMap
• New York City Open Data
• UK Road Safety Data
• SafeGraph Data

4 Analysing tools

Several tools can be employed for analyzing urban data, particularly focusing on geospatial and temporal data processing and modelling. Selected Python packages and software are provided for your reference:

• Geospatial analysis: Geopandas, QGIS.
• Temporal analysis: skforecast, PyCaret.
• Urban mobility: scikit-mobility, Trackubtel.
• Machine learning: Scikit-learn, pytorch, keras, tensorflow.

5 Project management

It is highly recommended to utilise GitHub for project management and code writing with version control. Further guidance on GitHub usage can be found at Github Docs. You can also find some online courses at Linkedin Learning, Udemy or Coursera. The simplest method involves utilizing GitHub Desktop to commit and push your local Jupyter Notebook project.

Appendix

Table A1 The information on current research topics

No	Title	Description
1	Geospatial analysis for urban crimes	Geospatial analysis for urban crimes can help to develop tailored place management strategies to prevent potential crime in urban areas. This project focuses on detecting crime patterns (e.g., crime hotspots and concentration) or exploring how urban socio-economic and environmental factors influence crime patterns in urban areas approached by geospatial analysis and techniques, from statistical geospatial models to advanced machine learning models.
2	Spatio-temporal analysis for urban crimes	Comprehending spatial and temporal patterns of urban crime is essential not only for unravelling the mechanisms behind when and where crimes occur but also for gaining valuable insights to formulate targeted intervention strategies. Spatio-temporal analysis for urban crimes aims to employ advanced data science methods to investigate the spatial and temporal patterns of crimes in urban areas. In the context of urban complexity, it can help to identify specific crime patterns such as near-repeat victimisation in burglary (via spatiotemporal clustering) in neighbourhood areas.
3	Crime prediction using machine learning/ deep learning	This project aims to develop an advanced crime prediction framework/method by leveraging advanced machine learning and deep learning techniques. By integrating historical crime data with relevant socio-economic and environmental factors, the project seeks to enhance the accuracy and efficiency of crime prediction in space and time.  The implementation will be designed for existing law enforcement systems to improve effective public safety.
4	Causal inference for crime pattern shifting	This project aims to delve into causal inference with a specific focus on understanding and addressing crime pattern shifting or replacement across urban areas. Crime patterns can undergo shifts due to various factors such as changes in law enforcement interventions (e.g., policing patrolling), and disruption of social conditions (e.g., pandemic, natural hazards). By revealing the interactional associations between neighbourhood characteristics and the resulting shifts/replacements in crime patterns via causal inference model, it can provide insightful intervention assessments/strategies for public policy.
5	Exploring the urban mobility patterns using big data analytics	Understanding and optimising mobility are crucial for sustainable and efficient modern urban development. The objective of this project is to explore the complex patterns in urban mobility by analysing diverse big data sources, such as public transportation records (e.g., smart card data), social media big data and mobile phone big data. Other tasks of this project can focus on identifying what key factors (urban facilities and functional land use) influence the population’s mobility patterns (e.g., commuting behaviours), or predicting the volume of the population’s mobility trends (e.g., origin and destination flows) across different urban areas. Employing alternative methods such as machine learning and geospatial analysis will be pivotal in extracting meaningful insights for urban planning or public resource management.
6	Exploring urban inequality using big data analytics	The project aims to reveal the multifaceted dimensions of urban inequality through the lens of advanced big data analytics. With the development of advanced geospatial analytic techniques, it is possible to utilise emerged big data sets to explore various inequality topics in the urban context, such as the inequality of residents’ travelling behaviours across different neighbourhood areas, or the accessibility of greenspace of residents across neighbourhood areas. The main task focuses on uncovering the spatial distribution of disparities across UK city areas (or region areas) and underlying the factors contributing to specific types of urban inequality (for building the prediction model).  To achieve these goals, alternative methods such as machine learning algorithms, and spatial network/graph analysis will be explored to enhance the accuracy of the findings, ensuring a nuanced examination of urban inequality in contemporary societies.
7	Urban transport analytics	The project aims to analyse the heterogeneity in urban transportation demand, usage/ridership, or model choices across urban areas to understand the diverse patterns of mobility within urban settings. By employing advanced geospatial data analytics and machine learning techniques, different types of transport mode patterns (e.g., taxi, cycling, public transportation) can be sensed, visualised, and analysed from various geo big data, such as smart card data, bike-sharing docking station data and mobile phone data.
8	Urban traffic accident analytics	This project aims to analyse and predict the traffic accidents (car, bike or pedestrian involved) within urban areas. With the increasing complexities of modern urban transportation networks, this project seeks to employ advanced AI techniques to extract meaningful insights from vast datasets related to traffic accidents. The primary task is to identify patterns, trends, and contributing factors (e.g., road features) leading to traffic accidents, thereby facilitating evidence-based decision-making for city planners. As an alternative to the traditional method, this project focuses on the integration of explainable machine learning and geospatial analysis to provide a comprehensive and dynamic approach to accident prevention and contribute to a safer urban environment.
9	Evaluating urban vitality /vibrancy using geo-big data	This project aims to evaluate the urban vitality across retail areas (or high streets) through the footfall traffic sensed from geo-big data. The primary task of this project is to explore the daily rhythms of vitality (represented by footfall traffic at place venues) across different urban land use areas. Second, it seeks to identify key factors influencing the vitality, encompassing aspects such as economic revitalization, social cohesion, and environmental sustainability. To achieve this objective, alternative research methods, including spatial and temporal analyses, and machine learning models will be explored to provide a comprehensive understanding of the dynamics of urban areas.
10	Sensing urban functions through big data analysis	In the context of dynamic urban environments, this project endeavours to employ AI tools to sense and comprehend various urban functions from geo big data.  Against the backdrop of rapidly evolving cities, understanding the intricate interplay of urban population interacting with diverse functions in the urban landscape is crucial for effective urban planning and management. The primary objective of this project is to detect and portray the dynamic urban function zones from human activity patterns sensed from geo big data (e.g., social media data, mobile phone data, smart card data, street view data and remote sensing data).

---

© 2024 Tongxin Chen. All rights reserved.

Contact: Tongxin.Chen@hull.ac.uk