Did COVID-19 Reshape Visitor Preferences in Urban Parks? Investigating Influences on Sentiments in Shanghai, China

Siqi Lai, Yifan Zhu, Brian DealSustainability 2023 · DOI

TL;DR

This study investigates how the COVID-19 pandemic has reshaped visitor preferences for urban parks in Shanghai, China. Using social media data and sentiment analysis, the research reveals a significant shift towards appreciating neighborhood parks and features like walking accessibility and proximity to amenities, offering crucial insights for future urban park design and planning.

What Problem Does This Paper Address?

Urban parks are vital for city dwellers' well-being, offering spaces for recreation, health, and social connection. The COVID-19 pandemic, with its lockdowns and social distancing measures, fundamentally altered how people interact with and perceive urban green spaces. While research has touched upon increased park usage during the pandemic, a critical gap exists in understanding how specific park attributes and environmental factors influence visitor sentiment in the post-pandemic era. This paper aims to fill this gap by examining whether the pandemic has changed preferences for urban parks and identifying the key features that drive positive visitor experiences.

Methods

The researchers employed a mixed-methods approach, combining sentiment analysis of social media data with quantitative modeling.

  • Sentiment Analysis: They utilized Long Short-Term Memory (LSTM) neural networks, a type of recurrent neural network adept at processing sequential text data, to analyze Sina Weibo posts related to urban parks. This method captures nuanced emotional tones and attitudes. To ensure accuracy, they validated their sentiment classification using two prominent Chinese NLP platforms (Tencent and Baidu), achieving high model accuracy (95% and 92%). The "Positive Sentiment Ratio" was calculated as the proportion of positive emotion posts to the total number of posts.
  • Quantitative Modeling: A Pearson correlation analysis was performed to identify relationships between park attributes and positive visitor sentiment. Subsequently, a random forest regression model was used to analyze the nonlinear relationships between these factors and sentiment.

Datasets

  • Social Media Data: Geo-referenced Sina Weibo posts containing park geotags were collected for the period July 2018 to March 2022. After cleaning and filtering, approximately 39,731 posts were used.
  • Park Inventory Data: Locational IDs for urban parks were obtained from the Shanghai Landscaping and City Appearance Administrative Bureau.
  • Land Use Data: 10m spatial resolution land cover data (forests, grasslands, water bodies, built areas) were sourced from Esri.
  • Points of Interest (POIs): Data on dining establishments, transportation hubs, and retail outlets were gathered from Baidu Maps.
  • Road Data: Open Street Map data was used to reflect park accessibility.
  • Household Prices and Resident Data: Information on house prices and resident population density was obtained from Lianjia.
  • Validation Data: Ratings from Dazhongdianping and Ctrip platforms were used to validate the social media sentiment analysis results.

Key Insights and Ideas

  • Neighborhood Parks Gain Prominence: The pandemic significantly boosted the appeal of neighborhood parks, which are closer to residential areas and became crucial for daily activities during lockdowns. This shift is a lasting consequence, not just a temporary trend.
  • Accessibility is King (Post-Pandemic): While water bodies remain important, walking accessibility and proximity to shopping facilities emerged as significantly more influential factors for positive sentiment after COVID-19. This highlights a desire for convenience and integrated lifestyles.
  • Nature's Enduring Appeal: Despite the rise of neighborhood parks, nature parks and historical gardens maintained their appeal, suggesting a continued appreciation for natural beauty and cultural heritage.
  • Sentiment Variability Matters: Neighborhood parks, while gaining popularity, showed the widest range of sentiment scores pre-pandemic, indicating a high degree of heterogeneity in their quality. Post-pandemic, this variability decreased, suggesting a more consistent positive experience.

Key Results

  • Pre-COVID-19: Nature parks generally received the highest positive sentiment ratios, followed by historical and comprehensive parks. Neighborhood parks had the lowest median score but the widest range of sentiment.
  • Post-COVID-19: A statistically significant increase in positive sentiment ratios was observed for neighborhood parks. Other park types did not show significant changes.
  • Key Influencing Factors (Pre-COVID-19): The ratio of water bodies, surrounding resident density, walking accessibility, walking path density, and surrounding shopping facility density were significant positive correlates of visitor sentiment.
  • Key Influencing Factors (Post-COVID-19): The importance of walking accessibility and surrounding shopping facility density significantly increased. Walking path density also saw a substantial rise in its correlation. The ratio of water bodies remained strongly correlated, though its relative importance slightly decreased.

Limitations and Future Work

  • Social Media Bias: The sentiment analysis relied on Sina Weibo data, which may not fully represent all user demographics or sentiments, as it excludes users who do not use the platform. Future research could incorporate additional data sources or mixed methods for a more comprehensive view.
  • Geographic Specificity: The study was conducted solely in Shanghai, limiting the generalizability of findings to other urban contexts globally. Comparative studies across different cities are recommended.
  • Long-Term Trends: While the study captures post-pandemic shifts, further investigation is needed to determine if these trends persist, intensify, or reverse over time as societies fully emerge from the pandemic's immediate effects.
  • Unconsidered Attributes: The study identified key attributes but acknowledges that other factors like park maintenance, safety, and programming could also influence visitor sentiment and warrant future exploration.

Summary updated

Abstract

Urban parks can be critical components of city landscapes. In the wake of COVID-19, understanding the role of urban parks in helping to elicit positive sentiment and improve the overall well-being of visitors has gained new traction in the literature. This research distinctively investigates whether the COVID-19 pandemic altered preferences regarding urban parks and identifies the key landscape attributes and environmental factors that influenced positive visitor sentiment, thereby addressing a critical gap in understanding the evolving dynamics of urban green spaces in the post-pandemic era. We use a mixed methods approach that includes natural language processing techniques to analyze crowd sourced data across more than 100 urban parks in Shanghai, China. Not surprisingly, our results highlight a post-pandemic rise in preferences for neighborhood parks and features typically associated with neighborhood parks, such as walking accessibility and surrounding residential densities. In addition, we found six park features, the presence of grasslands, water bodies, walking paths, and proximity to shopping facilities, significantly impacted the ways in which people perceived their park experience. These insights can guide urban park planning, design, and management in our evolving post-pandemic landscape to help ensure that urban parks continue to serve as essential urban spaces.
BibTeX
@inproceedings{lai2023did,
  author = {Siqi Lai and Yifan Zhu and Brian Deal},
  title = {{Did COVID-19 Reshape Visitor Preferences in Urban Parks? Investigating Influences on Sentiments in Shanghai, China}},
  booktitle = {Sustainability},
  year = {2023},
  doi = {10.3390/su152316396},
}
Figure