Document Type : Articles extracted from Thesis
Authors
Department of Human Geography and Planning, Faculty of Geography, University of Tehran, Tehran, Iran
10.22034/jsc.2024.427202.1748
Abstract
A B S T R A C T
Nowadays, one of the main challenges of big cities is the urban worn-out texture. Livability, as one of the important indicators in examining these textures' condition, can help identify their needs. The current research aims to use machine learning modeling methods to analyze the data related to the livability of urban worn-out textures in District 7 of Tehran, including the discovery of patterns and structures in the answers to the questionnaire and their prediction and evaluation. The research is applied in terms of purpose and descriptive-survey in terms of nature and method. The statistical population in this research is the residents of worn-out neighborhoods of District 7 of Tehran, with a number of 190,228 thousand people, and the statistical sample size is 383 people, who were selected by simple random method as a probability sampling. The research data were analyzed using two algorithms as random forest regression and k-means. The results showed that using a random forest regression algorithm, the livability level of eight neighborhoods was predicted, with the Gorgan neighborhood having the highest score and the Shahid neighborhood having the lowest livability score. Also, by using this algorithm, according to the questionnaire questions, it was determined that factors such as access to public transportation and health centers were in the highest ranks. The k-means algorithm divided the questionnaire data into five clusters based on the similarities in the answers to identify different patterns and needs of the residents. In this algorithm, cluster one, with 24% of respondents, showed high dissatisfaction with the quality of the urban environment and public health.
Extended Abstract
Introduction
Urbanization is increasing rapidly around the world, and it is predicted that by 2050, 68% of the world's population will live in cities. This population growth is associated with challenges related to urban development and excessive expansion of cities. The main focus of urban development should be improving people's quality of life in cities. In this regard, the concept of livability has been proposed as one of the latest theories in urban planning, which refers to people's satisfaction with environmental conditions and interactions and can be investigated at the scale of the city or its smaller parts, such as worn-out neighborhoods. The worn-out texture is one of the problems of Iranian cities, which has become inefficient due to the passage of time or lack of proper planning, and its consequence is the reduction of living conditions, safety, and physical, social, and economic disorders. District 7 of Tehran has more than 300 hectares of worn-out texture, mainly located east of this area. These textures face problems such as insufficient per capita service users, poor quality of buildings, and narrow alleys. In this research, using machine learning techniques, we seek to extract important features from the questionnaire data on the urban livability of worn-out textures in District 7 of Tehran and identify effective indicators to identify the initiatives that will significantly improve livability.
Methodology
The current research is applied in terms of purpose and descriptive-survey in terms of nature and method. The statistical population is all the residents of the blocks of urban worn-out texture in District 7 of Tehran, with a population of 190,228 people. Sampling was done in a simple random and probabilistic manner, and the minimum sample size was estimated to be 383 questionnaires using Cochran's formula. Documentary and field methods were used to collect information. A questionnaire was designed using the Delphi method with 5 dimensions, 22 components, and 158 sub-components, and its validity was confirmed with Cronbach's alpha of 0.857. After data collection and pre-processing, analysis was done based on machine learning models. A modeling pipeline was developed to analyze the data and extract insights related to livability drivers. The random forest regression model was used to predict neighborhoods' livability and the questions' importance. Also, an unsupervised k-means model was used to cluster residents based on common patterns of questionnaire responses.
Results and discussion
Based on the random forest regression model, Gorgan neighborhood has the highest livability rate in different dimensions with an average score of 5.281. Kaj, Khwaja Nasir, and Qasr neighborhoods are ranked second to fourth, respectively. The neighborhoods of Nizam Abad, Khaje Nizam, Armenians, and Shahid were ranked in the lowest levels of livability. The random forest model with R-squared equal to 0.62 matches the real data well. The random forest model has ranked the importance of questionnaire questions in predicting livability. The top ten features include the distribution of public transportation stations, access to medical centers, access to shopping areas, citizen responsibility, security, housing prices, environment cleanliness, traffic volume, educational space quality, and green space distribution. These factors are aligned with the main pillars of livability and show the importance of access to urban services, environmental conditions, security, educational facilities, and socio-economic conditions. Using the k-means clustering algorithm, the residents of the neighborhoods were grouped into 5 clusters with similar characteristics. The first cluster (24%) is from the low quality of the urban environment and public health, the second cluster (21%) from social welfare problems and quality of life, the third cluster (21%) from the lack of urban facilities and infrastructure, the fourth cluster (16%) from Inadequate welfare and economic situation, and the fifth cluster (18%) are dissatisfied with the problems of development and governance. Some of the results of this research are consistent with research findings that have used machine learning methods to assess livability. Other studies have also reported the difference in the livability level of different neighborhoods. The key role of access to public transportation, medical services, and shopping facilities in determining livability is consistent with the findings of other researchers. The results of this study emphasize the importance of citizens' participation and responsibility in improving livability, which is in line with the recommendations of previous studies. The most important innovations of this research include the use of advanced machine learning algorithms such as random forest and K-means, quantitative prediction of the livability of each neighborhood, extracting useful and interpretable insights from the model results, identifying different groups of residents with different needs, and combining the subjective data of the questionnaire with observations. The methodology used in this research can be generalized to other contexts and urban areas with adjustments.
Conclusion
The random forest regression algorithm showed that Gorgan and Shahid neighborhoods got the highest and lowest livability scores, respectively. Factors such as access to public transportation stations and health centers were the most important. The k-means algorithm classified the data into five clusters, and cluster one, including 24% of the respondents, showed high dissatisfaction with the quality of the urban environment and public health. Proposals for the reforms of each neighborhood include Gorgan, supervision of constructions, renovation of dilapidated buildings, creation of parks and recreational spaces; Kaj, continuous communication between people and officials, participation in construction projects, improvement of urban furniture, creation of tarebar market, playground and parking lot; Khajeh Nasir, holding entrance exam classes, creating green and recreational spaces to prevent the gathering of drug addicts; Qasr, skill training and job creation for female heads of the household, creation of large shopping centers; Nizamabad, developing a culture of participation, improving access to services, organizing incompatible businesses, creating parking, removing vermin; Khawaja Nizam al-Molk, widening the roads, creating a parking lot and a small park, supervising medical offices, health education; Armenians, improving the quality of roads, fixing traffic accidents, creating sports complexes, paying attention to green spaces; Shahed, increasing police patrols, installing CCTV cameras, improving street lighting, raising awareness for crime prevention.
Funding
There is no funding support.
Authors’ Contribution
The authors had equal contributions in all stages and sections of conducting the research.
Conflict of Interest
The authors declare that they have no conflict of interest regarding the authorship or publication of this article.
Acknowledgments
The authors would like to thank all those who assisted us in conducting this research, especially those who performed the quality assessment of the articles.
Keywords
