การวิเคราะห์ความสามารถเดินเท้าเพื่อเข้าถึงสวนสาธารณะภายในหมู่บ้านจัดสรร

Abstract

Public parks and green space are critical resources for enabling, encouraging, and promoting physical, mental, and social health to urban residents. Considering the strong association of residential proximity and physical activity, public parks located far away from the residential community are likely to be underutilized, deteriorated, and then abandoned, leaving the community at risk of becoming unlivable. This research therefore aims to study the influence of various factors, especially the walking distance, on the residents’ decision whether to use the public park in their housing estates. Comparative applications of two analytical models, Artificial Neural Network (ANN) and Multinomial Logistic Regression (MLR), were also investigated, of which the results can provide the guidelines for selecting the public park location. Given that, questionnaire surveys were conducted among housing estates and residential communities in Chiang Mai’s principle city area to investigate residents’ socio-economic data (e.g., income, education, health insurance, marital status, family structure, etc.), physical characteristics (e.g., age, gender, weight, height, etc.), traveling data (e.g., vehicle in possession, frequently used transport mode, daily travel patterns, etc.) and traveling attitudes (e.g., acceptable walking distance, etc.). After excluding incomplete questionnaires, data of 554 respondents were analyzed using ANN and MLR models. The results of both models correspondingly point out that there were four factors negatively influencing the respondents’ decision to walk to the park, including walking distance, health insurance, income, and age range. In addition, due to the model nature, each run of the MLR models consistently produces, with a linear relationship assumption, the same analytical results. On the other hand, the ANN models, without any restricted relationship assumption of the input variables, can yield different analytical results while generating a better learning outcome for each run. Combining these two models not only provide insights into the importance of explanatory variables and their influences, but also accordingly suggest the guidelines for locating the public park. It is recommended that, to cover at least 50 percent of the residents, the farthest house should be located not more than 500 meters from the park for the residential area with elderly houses and not more than 450 meters from the park for the housing estates in general.