Historical Review of Typological Evolution and Design Strategy Preferences of High-FAR Primary and Secondary Schools: Evidence from 67 Cases in Shenzhen

Abstract

Rapid urbanization has intensified the shortage of school places in many developing countries, prompting the rise of compact, high-floor area ratio (FAR) school models. However, research on high-FAR school design strategies remains limited. This study systematically analyzes 67 high-FAR schools in Shenzhen, China. Using design descriptions as the sample, the analysis applied the N-gram model and identified five major design strategies: responses to regulations, functional integration of classroom spaces, functional integration of public spaces, climate adaptation and sustainability, and alleviation of psychological stress. Correlation analysis revealed that factors including FAR, total floor area, design year of the schools, regional GDP and investment in the education sector significantly influence preferences for different design strategies. Further, K-means clustering categorized four types based on strategy adoption and FAR: the comprehensive strategy type; the user-centered innovation type; the spatial integration type; the psychological well-being type. The results emphasize the need for adaptable design strategies that reflect local development stages. These findings contribute to a data-informed foundation for improving spatial efficiency in rapidly urbanizing settings, offering policy and design guidance for rapid developing cities.

1. Introduction

The rapid urbanization and significant population growth since 1950 have contributed to increased urban congestion and a shortage of available land in many cities, approximately 55% of the global population resides in urban areas [1]. By 2030, it is projected that over 56% of people in developing countries will live in cities, with the majority of the urban population growth expected to occur in Asia and Africa [2]. Taking China’s first-tier cities as an example, by 2024, Shenzhen reached a population density of 8867 people per square kilometer. Its land development intensity has risen to approximately 50%. Among this, educational facilities occupy 1.48 square kilometers, accounting for 51% of the planned public service land area [3,4]. This trend suggests that cities and megacities worldwide must consider the scarcity of land resources and improve the efficiency of land use. This shift is particularly evident in economically advanced cities. For example, London plans to add 3000 new housing units each year to meet rising demand [5]; in Singapore, the General Redevelopment Area (GRE) value exceeds 2.8, officially qualifying it as a high-density zone [6]; in Hong Kong, the FAR in many residential developments within new towns typically ranges from 5.0 to 6.0 [7].

As a result, the rise of high-density cities poses challenges for the effective planning and allocation of public facilities, such as schools, hospitals, and sports centers, to ensure social equity [8]. Among these, school infrastructure is particularly critical, as it is widely discussed across various countries and directly influences both the social fabric and the long-term economic development of the city [9]. Most of these discussions focus on the macro level, such as school planning, connections with communities, and the design of specific public spaces and sports fields. For example, in Australia, a child-friendly vertical school environment is identified that should include expansive green terraces, open hallways, and other spaces to support children’s developmental needs. There are also studies that explore the potential of vertical schools as shared community resources from the perspective of spatial organization [10,11]. In Singapore, the design strategies of new generation school buildings have evolved according to local policies and learning habits, emphasizing that school layout and spatial atmosphere should meet the needs of modern education. The use of vertical sports fields as part of urban facilities is also discussed in terms of economic benefits [12,13]. In the UK, the design strategies for high-FAR schools, which typically feature buildings with more than six floors and a wider, lower profile, are shaped by the urban development processes of their respective cities. In high-density areas, schools often share activity spaces with apartment residents or use nearby community centers to solve the problem of limited school land [14]. In Italy, urban planners combine school spaces with city design strategies to activate urban vitality [15]. In Japan, planning and safety management are improved by combining sports fields and public facilities to create shared spaces for both schools and communities [16]. In South Korea, school public spaces are designed for multiple uses to serve the community, while also considering the psychological characteristics of different age groups and organizing school spaces accordingly [17].

Despite the global trend of declining population growth, urban populations and the demand for primary and secondary school placements continue to increase [18]. As a Special Economic Zone (SEZ) in China, Shenzhen has rapidly attracted migrants, particularly young individuals, drawn by economic and educational opportunities [19]. This influx of youth contributes to the ongoing rise in both the total population and the school-age population in the city.

According to data from the Shenzhen Statistical Yearbook, Shenzhen covers an area of 1997 square kilometers and has a population exceeding 17.68 million, resulting in the highest population density in China—approximately twice that of Shanghai, 1.2 times that of Singapore, and twice that of Tokyo. By 2024, Shenzhen planned to renovate and expand 160 primary and secondary schools, increasing the total number of available school places by 0.18 million [20]. By 2035, Shenzhen plans to provide 3.64 million basic education places, with at least 3 million new places expected to be built [21]. Given Shenzhen’s relatively recent urban development spanning just four decades, the construction of primary and secondary schools has been slow and insufficient in number, failing to keep pace with the city’s rapid population growth. Therefore, Shenzhen began developing primary and secondary schools in early 2012, along with proactive modifications to the design regulations for these schools. The construction of primary and secondary schools, exemplified by those in Shenzhen, has indirectly led to the revision of national design standards of school.

Table 1 shows the significant changes in Shenzhen’s regulations on primary and secondary school design and their impact on the design process. In 2012, China abolished the regulation that “the plot ratio of elementary schools should not exceed 0.8, and that of secondary schools should not exceed 0.9″ [22,23]. From then on, more high-density cities attempted designing and building schools where the FAR exceeds 0.9, and Shenzhen is the most radical. We define such schools, whose FAR exceeds the traditional 0.9 limit, as high-FAR schools. In 2018, Shenzhen launched the “Futian New Campus Action Plan”, aimed at addressing the educational resource constraints in Futian District and meeting the growing student population [24]. In 2019, Shenzhen districts issued guidelines to improve construction standards for primary and secondary schools, tailoring design recommendations to the specific needs and conditions of each district [25].

Table 1. Significant regulations affecting the design of High-FAR primary and secondary school design in Shenzhen.

Government Measures	Time	Impact on Design Process
Abolishment of code for design of school (GBJ99-86)	2012	Removed FAR constraints on school development
Initiation of Futian new campus plan	2018	Triggered a wave of high-FAR school construction
Release of upgraded school construction guidelines	2019	Provided greater flexibility in school facility development

Along with more than a decade of development, the construction of high-FAR primary and secondary schools in Shenzhen has gradually developed distinct characteristics and preferences such as compact design. Due to spatial limitations, high-FAR primary and secondary schools often lack sufficient land, which restricts the construction of activity areas, sports fields, and classrooms. A main feature is the vertical stacking of large public spaces such as sports fields and libraries. Among these, rooftop sports fields and underground activity areas are the most common. This approach can also improve the efficiency of space use and help break the limitation of single-function spaces, such as adding corridors as activity areas or achieving space sharing through layout optimization [26,27,28,29,30]. In the past, Chinese scholars and designers have typically presented school designs as individual case studies [27,29] and development reviews [30]. However, as a product of architectural design driven by special needs, high-FAR primary and secondary schools in China have received little systematic analysis regarding their development trends, common characteristics, and construction types. Therefore, this study aims to address four key research questions:

• Which high-FAR design strategies have been frequently used by designers over the past 13 years?
• With policy changes, what are the trends in high-FAR primary and secondary school design strategies from 2012 to 2025?
• What does the relationship between high-FAR design strategies in Shenzhen and contextual factors reveal about the underlying developmental mechanisms?
• What are the key design preferences of high-FAR school prototypes derived from the patterns of strategy adoption and FAR in Shenzhen?

By identifying these design strategy trends in Shenzhen, this study aims to present China’s perspective on high-FAR school design and rapid urbanization.

2. Methods

This study employs textual big data and mixed methods research approach that combines both qualitative and quantitative methods. Compared to traditional methods such as questionnaires and diagnostic interviews, collecting and analyzing data from big data and internet text is more efficient [31]. It also overcomes the issues of small sample sizes, limited time frames, incomplete coverage, and subjective data typical of traditional research methods [32]. In recent years, textual big data analysis methods have gained prominence in architecture-related fields [33,34,35,36,37]. To achieve the objectives of this study, the following four steps were carried out, as shown in Figure 1.

Figure 1. Research process flowchart.

Step 1: Select case. Figure 2a shows the location of Shenzhen in China, and Figure 2b shows the 67 cases selected for this study with a FAR exceeding 0.9 and construction dates after 2012, representing the majority of Shenzhen cases with detailed design descriptions currently available. These sources included official design websites, public accounts, and articles by the design teams. The design statements were prepared by project teams during the design and construction process and released through authoritative platforms, directly reflecting the core intentions and strategies, thereby serving as reliable textual evidence of current design trends in high-FAR school projects. Descriptions from authorized websites (e.g., Gooood, ArchDaily), which compile information from designer interviews and official sources, were also included, thereby enhancing the credibility of the data. Additionally, research cases included schools from most of Shenzhen’s administrative districts (Futian, Guangming, Longgang, Longhua, Luohu, Nanshan, Bao’an, Pingshan, Yantian). Since the case analysis relies on detailed descriptions of schools, Dapeng District, which has the lowest population density in Shenzhen, has very few schools with a FAR over 0.9. Therefore, no detailed design descriptions of high-FAR schools were obtained, and the district was excluded.

Figure 2. Locations of Shenzhen and cases.

Step 2: Crawl data. This study employed the Python-based Scrapy framework to crawl the design descriptions of 67 schools. By cleaning the web content and processing the text to retain only Chinese characters–while removing English, punctuation, and advertisements–we ensured the accuracy of the text data. This process resulted in approximately 78,863 valid Chinese characters. To extract structured linguistic features from the text, an N-gram model was applied to the cleaned raw text. Bigrams (two-character units) and trigrams (three-character units) were constructed to capture local semantic context and to calculate the frequency of each N-gram in the corpus. This process resulted in 3019 two-character word frequencies, 1937 three-character word frequencies, and 1564 four-character word frequencies

Step 3: Generate design strategy themes. This step combined both machine learning and manual verification. Key themes were manually identified from previous literature. The first and second authors independently reviewed previous research, using a three-gram coding technique to classify and summarize themes, where open coding was used until core themes emerged concerning design strategies for high-FAR primary and secondary schools [38]. They then discussed the original themes in recurring ideas until they reached complete agreement to ensure trustworthiness. In the following stage, the textual descriptions were used as samples for verification. The process became increasingly iterative, with field data collection and analysis carried out simultaneously to provide a coherent framework. Finally, the numerous coded themes were grouped into major categories. To further improve the accuracy of the analysis, this study utilizes machine learning methods to optimize the N-gram features, primarily by applying the Word2Vec word embedding model for semantic clustering and reducing word frequency errors. Word2Vec, introduced by Mikolov et al., is a seminal technique in Natural Language Processing (NLP) for learning high-quality, distributed word representations, commonly known as word embeddings [39]. The training process for Word2Vec learns word meanings simply by looking at which words appear near each other. During training, the model starts with random number assignments for each word (vectors), then repeatedly adjusts these numbers based on prediction errors, similar to how a student improves by learning from mistakes. The standard softmax probability for a target word ${\mathrm{w}}_{\mathrm{O}}$ given an input word ${\mathrm{w}}_{\mathrm{I}}$ (in the case of Skip-gram) is:

$$\mathrm{P}\left({\mathrm{w}}_{\mathrm{O}}|{\mathrm{w}}_{\mathrm{I}}\right)={\displaystyle \frac{\exp \left({{\mathrm{v}}_{{\mathrm{w}}_{\mathrm{O}}}^{‘}}^{\mathrm{T}}{\mathrm{v}}_{{\mathrm{w}}_{\mathrm{I}}}\right)}{\sum _{\mathrm{j}=1}^{\mathrm{V}} \exp \left({{\mathrm{v}}_{{\mathrm{w}}_{\mathrm{j}}}^{‘}}^{\mathrm{T}}{\mathrm{v}}_{{\mathrm{w}}_{\mathrm{I}}}\right)}}$$

(1)

where

v_w is the input vector representation of word w.

${\mathrm{v}}_{\mathrm{w}}^{‘}$ is the output vector representation of word w.

W is vocabulary size

The dimensionality reduction happens naturally because instead of representing each word as a huge sparse vector with thousands of dimensions (one for each word in vocabulary), Word2Vec compresses this into a dense vector of typically 100-300 numbers, capturing the word’s meaning in this compact form. The model ensures semantic consistency by continuously adjusting vectors so that words appearing in similar contexts (like “you xiao” and “gao xiao”, meaning efficient in Chinese pinyin) develop similar numerical representations, while maintaining format consistency through preprocessing steps like converting all text to lowercase and treating phrases like “Shenzhen school” as single units. The purpose of this process was to merge semantically similar theme words and to identify and analyze deeper design topics within the texts. The original themes were compared with the refined word frequencies, and they were manually refined. Terms and strategies were added, removed, or integrated. For example, the theme “design factors” lacked references to nature and material, so we updated the term, and included these aspects. Additionally, the “circulation optimization” term was absent in the crawled data, and consequently removed. Ultimately, we revised our initial categorization, refining six themes and 26 strategies into 5 new themes and 18 design strategies.

Step 4: Analyze data. This study conducted the following analysis to reveal the development and preferences of high-FAR primary and secondary school design strategies in Shenzhen.

• Data analysis 1 conducted a historical review of the collected cases and summarized the frequency analysis of design strategies. To minimize errors caused by sample size discrepancies across conditions, we normalized the word frequency data by calculating the ratio of each term’s frequency to the total word frequency within each condition;
• Data analysis 2 conducted a correlation analysis between social factors, design assignments, and design strategies. Eight relevant factors were identified from previous studies: FAR, total floor area, number of classes, design year, investment in Shenzhen’s education sector, Shenzhen’s GDP, number of schools, and population density across districts [40,41,42,43]. Among the factors related to social factors (design year, investment in Shenzhen’s education sector, Shenzhen’s GDP, number of schools, and population density across districts), all data were collected based on the year each case was designed, in order to ensure temporal consistency across variables. Three sets of variables were analyzed: five design strategy themes were considered dependent variables, while design assignment factors and social factors were considered independent variables, the processed word frequency data were imported into SPSS for correlation analysis. Before the analysis, the variables were tested for normality. The results indicated that the FAR, design year, investment in Shenzhen’s education sector, population density across districts, and Shenzhen’s GDP followed a normal distribution, allowing the use of Pearson correlation analysis. In contrast, the total floor area, number of classes, and number of schools did not follow a normal distribution, so Spearman correlation analysis was applied;
• Data analysis 3 conducted a clustering analysis on current high-FAR primary and secondary school cases in Shenzhen. The 67 collected cases were analyzed using the K-means clustering method, with the frequency of design strategies and the FAR as quantitative variables. The high-dimensional features were reduced to a two-dimensional space using Principal Component Analysis (PCA) for visualization. PCA1 and PCA2 represent the primary and secondary directions of variance, respectively, and are used in the plot to illustrate the distribution characteristics of different clusters, resulting in four distinct groupings.

3. Results

3.1. Design Strategy and Tendency of High-FAR Primary and Secondary Schools

As shown in Table 2, five design strategy themes and eighteen design methods are frequently employed by architects to address high-FAR challenges, based on the textual data analysis. These five include: responses to regulations; functional integration of classroom spaces; functional integration of public spaces; climate adaptation and sustainability for Shenzhen; alleviating psychological stress in high-FAR environments.

Table 2. Results of design strategy themes and the design methods.

Design Strategy Themes	Categories	Design Methods
1. Responses to regulations	Exceeding regulatory requirements	Reducing building spacing
		Reducing building setbacks
		Locating auxiliary teaching spaces above five floors
	Complying with Regulations	Complying with fire safety regulations
		Complying with minimum classroom dimension
2. Functional integration of classroom spaces	Enhancing the efficiency of classroom spaces	Designing multi-functional spaces
		Enabling shared use across student groups
		Consolidating functions
3. Functional integration of public spaces	Enhancing the efficiency of public spaces	Utilizing rooftop spaces
		Expanding underground spaces
		Enhancing the usability of transitional spaces
4. Climate adaptation and sustainability for Shenzhen	Ventilation	Elevated ground floor
		Incorporating double-height courtyards
	Shading	Using façade shading devices
		Implementing vertical greening
5. Alleviating psychological stress in high-FAR environments	Psychological well-being	Using psychologically supportive materials
		Using bright colors
		Incorporating natural elements

3.1.1. Responses to Regulations

The construction of high-FAR primary and secondary schools is largely constrained by regulations. As a result, many high-FAR schools have begun to either comply with or challenge these regulations. Most schools create additional functional spaces by adhering the minimum evacuation distance required by regulations; other strategies breaking the limitations of traditional regulations include: allowing non-primary teaching spaces above five floors, using soundproof materials to reduce building spacing, and negotiating with local governments to exceed regulatory setbacks [44,45]. For example, Xinsha Primary School [46] was granted a special setback approval by the Planning Department to allow the building’s podium to reduce the original setback from 6 m to 3 m (see Figure 3a). This adjustment provided the school with additional land for construction and more space for school activities. Furthermore, the area beneath the podium was made available to the local community, significantly enhancing the surrounding urban environment.

Figure 3. High-FAR school design strategies.

3.1.2. Functional Integration of Classroom Spaces

High-FAR conditions result in limited space availability for schools, thereby necessitating more integrated designs to improve classroom utilization efficiency. Through flexible design and pre-installed facilities, classrooms can be adapted to accommodate various types of courses. Alternatively, shared activity areas can be used to concentrate space and reduce transitional zones between classrooms. For example, Hongling Experimental Primary School [47] features two adjacent teaching spaces, divided by movable partitions to facilitate independent teaching activities on a daily basis (see Figure 3b). When larger-scale teaching or events are needed, the partitions can be removed, allowing the space to expand and accommodate these activities.

3.1.3. Functional Integration of Public Spaces

Necessary design of vertical school is an effective solution to address the challenges of high-FAR development. Traditionally, large public spaces such as sports fields are arranged independently on the ground, requiring substantial land area. By adopting public spaces on the same floor or in a vertical stacked configuration—such as placing sports fields on rooftop levels or extending activity areas underground—land use efficiency can be significantly improved. In some cases, corridors are expanded and transformed from simple passageways into multifunctional spaces for rest or informal activities. For example, Lianhua Primary School [48] employed a vertical design that converted the roof into a sports field, incorporating both a running track and a basketball court despite limited site conditions (see Figure 3c). This approach not only utilizes space efficiently but also provides children with accessible outdoor activity areas, thereby enhancing their interaction with nature.

3.1.4. Climate and Sustainability

High-FAR compact designs typically require the sacrifice of courtyards and transitional spaces. However, distinct from other high-density cities in China, Shenzhen’s hot and humid Lingnan climate necessitates particular emphasis on ventilation and shading in schools to ensure thermal comfort. These essential spaces for improving environmental comfort cannot be neglected. To improve ventilation, the ground floor of school buildings is often elevated or designed with double-height courtyards. In terms of shading, architectural devices and vertical greening strategies are commonly employed to reduce solar radiation. For example, Shenzhen Fuqiang Elementary School [49] replaces traditional balustraded corridors with perforated panel enclosures (see Figure 3d). These sun shading elements help reduce solar heat gain and lower temperatures while allowing for adequate air circulation.

3.1.5. Alleviating Psychological Stress in High-FAR Environments

High-FAR and spatially intensive designs can potentially have adverse effects on children’s mental health. However, exposure to natural elements and the use of color in materials have been shown to reduce stress and anxiety while enhancing mood and overall well-being [50,51]. Strategies include the use of eco-friendly materials to minimize air pollution caused by rapid construction, and the integration of color and natural features to support students’ mental health. For example, Pingshan Innovation School [52] incorporates red terracotta tiles as a decorative element (see Figure 3e). The building’s red exterior, complemented by surrounding greenery, creates a natural aesthetic that fosters a warm and welcoming atmosphere, allowing students to feel immersed in a nature-inspired environment.

3.2. Statistics on the Historical Development Process

Figure 4a illustrates the results of statistical analysis 1, which indicates the adoption rates of five school design strategies across three categories of FAR.

Figure 4. Strategy changes along with FAR and years.

Based on previous studies, FAR greater than 1.7 was used as the baseline for higher-FAR classification, and further levels were defined using a doubled threshold. To reduce subjectivity in threshold selection, one-dimensional natural break analysis was applied to validate the data [53]. Analysis of 67 FAR values identified two breakpoints around 1.8 and 3.2. Sensitivity checks using alternative thresholds (1.6 and 3.2; 1.8 and 3.6) showed that most samples remained in the middle group (about 63–67%), while the high-FAR group accounted for 9–21%. Because FAR 3.4 is slightly higher than the second breakpoint and therefore stricter, it was adopted as a practical standard for ultra-high FAR. Based on this criterion, the 67 collected cases were categorized into three groups: high-FAR (FAR < 1.7), higher-FAR (1.7 ≤ FAR ≤ 3.4), and ultra-high-FAR (FAR > 3.4).

The horizontal axis represents the FAR, while the vertical axis shows the proportion of schools adopting each strategy within each group. The chart reveals that strategy 1 (responses to regulations) has the highest adoption rate in the high-FAR school (FAR under 1.7) and shows a clear declining trend as FAR increases. In contrast, the adoption rates of Strategies 2 (functional integration of classroom spaces), 3 (functional integration of public spaces), 4 (climate adaptation and sustainability for Shenzhen), and 5 (alleviating psychological stress in high-density environments) have all increased, with Strategy 4 showing the most significant rise—from 22% to 31%. This suggests that schools with a FAR under 1.7, design strategies tend to emphasize planning controls and regulatory requirements. However, as FAR increases, designers are more inclined to implement strategies focused on functional integration, climate responsiveness, and psychological well-being. It is also noteworthy that schools with a FAR exceeding 3.4, although the adoption rate of Strategy 5 slightly declines, Strategies 3 and 4 maintain relatively stable usage levels. This indicates that even in extremely compact spatial conditions, these two strategies remain highly adaptable and essential for maintaining the functionality and environmental quality of school spaces.

Figure 4b illustrates the results of statistical analysis 2, which indicates the adoption rates of five school design strategies across three design time periods. The analysis uses the main Shenzhen government measures of primary and secondary school design as the key time point [22,23,24,25]. The timeline is divided into three periods: before 2018, 2018–2019, and after 2019. After conducting word frequency analysis, data show that Strategy 1 (responses to regulations) had the highest adoption rate before 2018, exceeding 18%, but experienced a steady decline thereafter, reaching its lowest level after 2019. This shift reflects a decreasing reliance on traditional regulatory constraints over time, which can be attributed to high-FAR design strategies that have reduced government restrictions on school regulations. Meanwhile, although the adoption of Strategy 2 (functional integration of classroom spaces) declined, Strategies 3 (functional integration of public spaces) and 4 (climate adaptation and sustainability for Shenzhen) demonstrated an upward trend throughout the timeline. This reveals a shift in designers’ focus—from improving classroom efficiency toward exploring spatial innovation in public areas—while also emphasizing adaptation to Shenzhen’ s local climate conditions. Notably, after 2019, Strategy 5 (alleviating psychological stress in high-density environments) began to decline in adoption, with its rate dropping by more than 3%. This suggests that some current ultra-high-FAR school designs may still lack sufficient consideration for students’ psychological well-being.

3.3. Correlation Analysis of Design Assignment and Social Factor

3.3.1. Correlation Analysis of Design Factors

Table 3 shows the results of the correlation analysis between design assignments and strategies. The results indicate that the FAR is significantly and positively correlated with Strategy 1 (responses to regulations) (r = 0.359, p < 0.05), Strategy 3 (functional integration of public spaces) (r = 0.284, p < 0.01), and Strategy 5 (psychological stress alleviation) (r = 0.336, p < 0.05). The total floor area also shows significant positive correlations with Strategy 3 (r = 0.266, p < 0.01) and Strategy 4 (climate adaptation and sustainability for Shenzhen) (r = 0.301, p < 0.01). In contrast, the number of classes does not exhibit a significant correlation with any of the five design strategies. Although some correlation coefficients (|r| < 0.3) fall below the conventional threshold, they are classified as medium effects by Cohen [54]. In social and developmental research, effect sizes often range between 0.17 and 0.31 and are considered meaningful in complex systems [55]. Therefore, when interpreted within the disciplinary context and empirical distribution, our results provide valuable evidence of the general link between design strategies and social factors.

Table 3. Result of correlation analysis of design factors.

	1 Responses to Regulations	2 Functional Integration of Classroom Spaces	3 Functional Integration of Public Spaces	4 Climate Adaptation and Sustainability for Shenzhen	5 Alleviating Psychological Stress in High-FAR Environments
Factor a	0.359 **	0.060	0.284 *	0.120	0.336 **
Factor b	−0.242	−0.032	0.266 *	0.301 *	0.061
Factor c	−0.196	−0.008	−0.004	0.194	−0.067

Notes: Factor a refers to the FAR; Factor b refers to the total floor area; Factor c refers to the number of classes. * p < 0.05; ** p < 0.01.

These findings suggest that as FAR and total floor area increase, designers are more likely to prioritize considerations related to regulatory constraints, public space optimization, climate responsiveness, and students’ mental well-being. Such strategies are crucial for managing spatial efficiency and environmental quality under high-FAR conditions. Furthermore, the results reflect a growing awareness among designers of the potential psychological and behavioral impacts of spatial crowding, leading to the integration of environmental behavior principles to enhance spatial functionality and address the increasing complexity of contemporary educational design.

3.3.2. Correlation Analysis of Social Factors

Table 4 shows the results of the correlation analysis between social factors and strategies. The results indicate that Factor A (design year) is negatively correlated with Strategy 1 (responses to regulations) (r = −0.241, p < 0.01), while its correlations with the other strategies are weak and not statistically significant. Factor B (the investment in Shenzhen’s education sector) shows a significant negative correlation with Strategy 2 (functional integration of classroom spaces) (r = −0.271, p < 0.01) and significant positive correlations with Strategy 3 (functional integration of public spaces) (r = 0.268, p < 0.01) and Strategy 4 (climate adaptation and sustainability for Shenzhen) (r = 0.259, p < 0.01). Similarly, Factor C (Shenzhen’s GDP) is significantly negatively associated with Strategy 2 (functional integration of classroom spaces) (r = −0.254, p < 0.01) and positively correlated with Strategy 3 (functional integration of public spaces) (r = 0.273, p < 0.01). Factors D (the number of schools) and Factor E (the population density across districts) do not exhibit any statistically significant correlations with the five identified strategies, implying a limited or indirect influence on current high-FAR school design approaches.

Table 4. Result of correlation analysis of social factors.

	1 Responses to Regulations	2 Functional Integration of Classroom Spaces	3 Functional Integration of Public Spaces	4 Climate Adaptation and Sustainability for Shenzhen	5 Alleviating Psychological Stress in High-FAR Environments
Factor A	−0.241 *	−0.163	−0.148	−0.128	0.099
Factor B	−0.056	−0.271 *	0.268 *	0.259 *	0.105
Factor C	−0.077	−0.254 *	0.273 *	0.237	0.150
Factor D	0.219	0.072	0.088	−0.015	0.146
Factor E	0.237	−0.201	0.031	−0.021	0.142

Notes: Factor A refers to the design years; Factor B refers to the investment in Shenzhen’s education sector; Factor C refers to Shenzhen’s GDP; Factor D refers to the number of schools; Factor E refers to the population density across districts. * p < 0.05.

These findings suggest that, as primary and secondary school construction in Shenzhen has progressed, designers place relatively less emphasis on regulatory compliance and shift their focus to other design strategies. This pattern may be due to multiple factors. In high-density settings, regulation-oriented responses alone are less effective than space-efficiency-oriented approaches in dealing with land shortages and functional stacking, which can be time-consuming and require considerable effort. Additionally, some schools reduce setback space, which weakens buffer zones and increases potential safety risks. Results also indicate a contrasting trend between classroom function integration and public space integration strategies as government investment in education increases and regional economic conditions improve. With higher construction budgets, designers are no longer required to compress multiple functions into a single classroom. For example, musical instruments like pianos can be relocated from general classrooms to dedicated music rooms. At the same time, increased funding has enabled the development of multi-level layouts for sports facilities and large public areas, providing an effective solution to spatial constraints in high-FAR schools. In addition, increased financial investment has led to greater emphasis on comfort-oriented design in schools. For example, the use of more efficient vertical shading devices or the integration of vertical greenery on building facades can significantly improve thermal comfort in school spaces.

3.4. Cluster Analysis for Categorizing

Table 5 and Figure 5 present the result of the K-means clustering analysis, which produced four distinct categories of 67 high-FAR schools based on design strategy frequency. The average silhouette coefficient is 0.641, indicating good separation and compact clusters. The Davies–Bouldin index is 0.443, and the Calinski–Harabasz index is 278.803. These internal indices support the adequacy of the four-cluster solution. As shown in the images, high-FAR primary and secondary schools in Shenzhen can be categorized into the following four types: Comprehensive strategy type with high-FAR (Cluster 1), User-centered innovation type with ultra-high-FAR (Cluster 2), Spatial integration Type with higher-FAR (Cluster 3), and Psychological Well-being Type with higher-FAR (Cluster 4).

Table 5. Result of clustering evaluation indices.

Silhouette (Mean)	DBI	CH
0.641	0.443	278.803

Notes: DBI refers to the Davies–Bouldin index; CH refers to the Calinski–Harabasz index.

Figure 5. Clustering analysis yielded four distinct clusters.

Cluster 1: Comprehensive strategy type with high-FAR. As illustrated in Figure 6a, this cluster accounts for 11.94% of the cases, with FAR primarily ranging from 0.9 to 1.6. Projects in this group demonstrate balanced adoption of all five strategies, with a particularly strong emphasis on Strategy 1 (responses to regulations), representing a comprehensive and regulation-conscious design approach.

Figure 6. Radar chart of cluster analysis.

Cluster 2: User-centered innovation type with ultra-high-FAR. As illustrated in Figure 6b, this cluster accounts for 28.36% of the cases, with FAR primarily exceeding 3.2, corresponding to extreme FAR schools. These schools adopt minimal use of Strategy 1, while widely integrating Strategies 3 (functional integration of public spaces) and strategies 4 (climate adaptation and sustainability for Shenzhen), signaling a shift toward user-centered, integrated, and adaptive design models.

Cluster 3: Spatial integration type with higher-FAR. As illustrated in Figure 6c, this cluster accounts for 32.84% of the cases, with FAR primarily ranging from 1.6 to 3.2, corresponding to ultra-high FAR schools. Projects in this group emphasize Strategy 3 (functional integration of public spaces) and strategies 4 (climate adaptation and sustainability for Shenzhen), highlighting the use of vertical layouts and environmental strategies to optimize spatial efficiency.

Cluster 4: Psychological well-being type with higher-FAR. As illustrated in Figure 6d, this cluster accounts for 29.85% of the cases, with FAR primarily ranging from 1.6 to 3.2, corresponding to ultra-high FAR schools. These projects feature a high adoption of Strategy 5 (alleviating psychological stress in high-far environments), reflecting a strong focus on emotional comfort and mental health in dense educational settings.

The clustering analysis above shows that high-FAR schools and ultra-high-FAR schools exhibit strong design preferences. High-FAR schools tend to employ a broader range of design strategies, as compared to ultra-high-FAR schools, due to the greater flexibility in available space for choosing different design strategies. In contrast, ultra-high-FAR schools, due to their highly compact spaces, prioritize addressing the negative impacts that such dense environments may have on children. However, higher-FAR schools, in addition to favoring compact space design strategies to enhance space utilization, have resulted in two different categories regarding attention to children’s psychological well-being. These results suggest that some current high-FAR primary and secondary school designs focus solely on functionality and space, neglecting children’s emotional well-being. This is an important consideration for future school designs.

4. Discussion

4.1. Differences in Design Strategies Between China and Other Countries

The design strategies for high-FAR primary and secondary schools in China differ significantly from those in other countries. In Singapore, the planning encourages different schools to share the same gymnasium and outdoor sports field. In terms of facilities, while ensuring school safety, auditoriums, libraries, and music classrooms are made accessible to nearby educators and communities with close partnerships, as seen in the UK, where schools have integrated public spaces into the second-level platforms of high-rise apartments to share with apartment residents during non-teaching hours [10,11]. However, building setbacks serve as both a safety buffer for children and a privacy barrier for residents, so minimum standards must be met. In some cases, sports fields are even used as buffer zones. Classroom layout emphasizes flexibility, based on a modular unit of 90 square meters, which can be multiplied as needed. Classrooms are designed without fixed functions and use flexible partitions according to actual needs. Multiple standard classrooms are grouped to share a specialized classroom, improving space efficiency. This kind of functional adaptability focuses on flexible use rather than predetermined zoning [21].

In contrast, in Shenzhen, the vast majority of schools are government-funded, with a primary focus on autonomy. Schools adopt a closed management model to maintain separation from the surrounding urban environment. School design places greater emphasis on the functional independence of educational facilities, which is totally different from countries where schools are integrated with the surrounding community to promote shared use and efficient resource allocation. For example, in the Shenzhen School Park, where schools are densely located, each school still has its own independent sports field. In terms of community access, some schools open their sports fields to nearby residents during scheduled time periods. In cases where a site faces limitations in school space during renovation and is unable to provide an independent sports field, the school connects its buildings to a nearby open sports area via a bridge [56], achieving efficient integration of teaching and public resources. For libraries and similar facilities, access is not granted to the public due to safety concerns. Regarding building setbacks, some schools reduce the required distance to increase the available construction area. While flexible partitions are also used in Shenzhen’s classrooms to meet various space needs, they mainly serve different activity requirements within the same class, such as regular lessons and class events. This differs from Singapore’s approach, where classrooms are not assigned fixed functions from the beginning.

4.2. Interaction Between Policy Promotion and Design Feedback Mechanisms

In the context of high-density urban development, government policy and school design have gradually formed a mutually beneficial relationship. Taking Shenzhen as an example, faced with the increasingly limited supply of educational land, policies have gradually relaxed rigid requirements for school designs. Although regulations initially required sports activity areas to be located on the ground floor [23], successful cases of integrating rooftop sports fields and activity spaces into platforms have driven adjustments to related policies [57]. Additionally, the empirical evidence of the effectiveness of design strategies in addressing the psychological stress associated with high-FAR schools has led to the reasonable compression of the per-student area in Shenzhen’s general primary and secondary school construction standards, reducing it from the initial 2.1 square meters per student to the current 1.7 square meters.

This shift is guiding schools design in China from strict compliance to environmentally sensitive solutions. Schools are meeting basic safety and functional standards while actively responding to policy flexibility through technology and design strategies, thereby developing more adaptable and resilient design languages to accommodate increasingly complex construction environments.

4.3. Applicability and Conditional Innovation in High-FAR Design Strategies

According to the results of this study’s analysis of social factors, the design strategies of most high-FAR primary and secondary schools are closely linked to regional educational investment and economic development levels. Economic conditions exhibit a significant bidirectional driving effect on high-FAR design innovation: in situations with tight budgets, constraints imposed by investment pressure often stimulate designers to explore space performance, material control, and functional integration more deeply, resulting in more cost-effective, energy-efficient, and even eco-friendly design solutions. In regions with abundant financial resources, there is greater potential to promote vertical space integration, large-scale greening, and the systematic application of green technologies. Economic conditions not only influence the form of high-FAR design but also play a profound role in shaping the mechanisms of strategy generation.

Ultimately, high-FAR design strategies are a direct response to the spatial pressures caused by the scarcity of urban land resources. This strategy is highly adaptable in developed areas with scarce land and high educational demand, but if applied indiscriminately in areas with relatively abundant land resources or lower urban density, it could lead to resource wastage and operational burdens. It is especially important to note that continuous financial investment is not a long-term sustainable development mechanism. Therefore, innovative practices in high-FAR school design should not be confined to a single model. In the context of simultaneous fiscal constraints and diverse urban investment pressures, designers should adopt truly forward-thinking strategies based on the differences in urban development stages and investment capacities, which transcend the path dependence driven by financial constraints and focus on strategic innovation based on actual usage needs. This will ensure the efficient utilization and long-term development of school spaces within limited resources.

4.4. Strategy Recommendations Based on Different FAR

For high-FAR schools with FAR of 1.7 or lower and relatively abundant space, the design should fully comply with current regulations to ensure basic safety and functionality, while improving spatial quality through localized optimization. The design focus at this stage should be on regulatory compliance and classroom function integration strategies. For higher-FAR schools with FAR between 1.7 and 3.4, although space is limited, there is still a degree of flexibility, making this range the most diverse in terms of design strategies. It is recommended to prioritize strategies for public space integration, climate adaptation, and psychological care, using vertical and modular designs to meet multi-functional layered demands and enhance spatial experience. For ultra-high-FAR schools with FAR exceeding 3.4, where space is severely limited, the core strategies should prioritize public space integration and climate adaptation. The core strategy should focus on public space integration and climate adaptation. More importantly, for ultra-high-FAR schools, special attention should be given to the negative psychological impacts of the high-FAR environment on children. Designers should avoid concentrating all functions on lower floors and actively explore mixed-use layouts such as high-level activity spaces, rooftop playgrounds, and subterranean corridors.

4.5. Limitations

This study is based on textual data and includes the majority of high-FAR schools in Shenzhen that feature detailed design descriptions. However, some schools with FAR values exceeding 0.9 were excluded due to the absence of sufficiently descriptive texts, resulting in the lack of case data from regions such as Dapeng. We acknowledge that excluding these less-documented areas may introduce potential regional bias and affect the generalizability of our findings. Although the variance explained by our results is limited and most correlations are small to medium in magnitude, it is important to note that, in social and developmental research, effect sizes between 0.17 and 0.31 are often considered meaningful in complex systems. For example, FAR is positively correlated with Strategy 1 (responses to regulations, r = 0.36), Strategy 3 (functional integration of public spaces, r = 0.28), and Strategy 5 (alleviating psychological stress, r = 0.34), suggesting that FAR can explain approximately 10–20% of the variation in strategy adoption. While not deterministic, these effect sizes are non-negligible under the multi-constraint conditions of school design. Other influencing factors, such as site conditions, functional needs, and design preferences, also play a significant role. In practical terms, the design of a school is never the outcome of a single factor, but rather the result of multiple forces working in combination. Therefore, these influencing factors identified in this study should be interpreted as directional guidance rather than strong predictors. They offer valuable insights for designers navigating the choices and complexities of high-FAR school development.

5. Conclusions

Under rapid urbanization, intensive land use in megacities has become an imperative trend. Particularly in allocating basic education facilities, megacities must create more educational placements within limited land resources to accommodate the demands of population influx and growth. Shenzhen, with the highest population density in China, has made a new attempt by building a large number of high-FAR primary and secondary schools since 2012. Over the past decade, Shenzhen has accumulated extensive completed high-FAR school projects, yet scholarly literature rarely synthesizes their design strategies, evolutionary patterns, and typological characteristics. This study, through clustering and statistical analysis of 67 primary and secondary school cases in Shenzhen, identified design strategy preferences across different FAR ranges, and provided a scientific basis for strategic choices in the design of schools with varying FAR types. The results show that:

Over time, the reliance on regulation-oriented design strategies has continuously declined, with design priorities shifting toward spatial integration and innovative approaches to better address the challenges posed by high-density environments. In regions with higher levels of economic development, schools benefit from greater flexibility in selecting design strategies and can optimize spatial configurations using diverse resources. Conversely, in resource-constrained areas, limited conditions have often motivated designers to adopt integrated and cost-effective strategies, achieving spatial innovation through efficient use of space and functionality.

Integrating FAR with design strategies, current high-FAR schools can be clustered into four predominant typologies: comprehensive strategy type, user-centered innovation type, spatial integration type, and psychological well-being type. For comprehensive strategy type schools with a FAR under 1.7, where construction land is relatively sufficient, designers tend to adopt a variety of spatial strategies to accommodate the demands of high-density environments. For user-centered innovation type schools with a FAR exceeding 3.2, where construction land is limited, spatial efficiency is improved by enhancing the utilization of both classroom and public spaces. For spatial integration type schools with a FAR ranging from 1.7 to 3.2, when construction budgets are sufficient, design strategies often integrate spatial layout with climate-responsive considerations. However, for psychological well-being type schools within the same FAR range of 1.6 to 3.2, where construction budgets are constrained, greater emphasis is placed on addressing children’s emotional needs under compact spatial conditions.

This FAR-based stratified strategy provides strong decision-making guidance for design teams in different urban environments, preventing the homogenization of spatial solutions and better addressing design challenges under varying FAR conditions. Building on this, future research could further establish a data model centered on the relationship between design strategy preferences, functional combinations, and spatial constraints, forming a strategy database and visualization decision-making tool to provide a rapid response design strategy recommendation system for projects with varying FAR. This model could serve as a reference for generating early-stage plans for new projects and is also expected to play a role in intelligent design assistance platforms, enhancing the scientific and forward-looking aspects of educational space design.