全國中等學校運動會(簡稱全中運)是臺灣參賽人數最多的賽會,競賽成績達前八名者可獲得甄試升學的資格,因此,在縣市端及學校端均相當重視此賽會的成績。本研究旨在蒐集歷屆得牌資料及是否設有體育班,運用奧林匹克運動會獎牌預測的模型,進行分析金牌數量之預測因子,期能初步建構全中運賽會成績預測模型。研究採用次級資料分析法,以Python程式語言對2023年全中運官網的歷年成績進行網路爬蟲,共計下載40,043筆成績紀錄,再串接體育班資料,轉換成個人層次的60,291筆資料,並運用Tobit模型進行分析。結果顯示有11個縣市在獎牌數量呈現上升趨勢,兩個縣市呈現下降趨勢,其餘縣市則是持平的表現。在六都直轄市、體育班比例和選手過去累積金牌數量,是預測全中運金牌數量重要的關鍵因素。本研究結論驗證金牌數量與縣市綜合經濟實力息息相關,從數據亦可觀察到縣市端運動競技實力動能與變化趨勢,透過運動賽會得牌預測模型分析追蹤,可檢證各縣市的體育政策和培育績效,並為未來的體育政策與資源分配提供有力的數據及決策支持。
The National High School Games (NHSG) is the sports competition with the largest number of participants in Taiwan. Athletes who achieve a ranking within the top eight are eligible for certain college admissions considerations, which leads both counties/cities and schools to place significant emphasis on NHSG outcomes. This study conducted secondary data analysis by collecting data on NHSG medal results from the past ten years and information about the schools of medal-winning athletes, including whether they have sports talent classes. Utilizing the Tobit model, which is employed internationally for predicting Olympic medals, the study analyzes the predictive factors for gold medal counts across different counties and cities, then creates a model for NHSG competition performance prediction. A web crawler using the Python programming language was employed to extract a total of 40,043 records from the 2023 NHSG official website. After associating the acquired data with sports talent classes, the dataset was transformed into individual-level records, resulting in a total of 60,291 entries. The results revealed an upward trend in medal count in eleven counties/cities but a decline in two, while the remainder remained relatively stable. Factors such as the proportion of sports talent classes in the six major metropolitan areas and the previous gold medal count earned by athletes emerged as significant predictors for forecasting the trends in gold medal acquisition. We conclude that a clear connection between the number of gold medals and the overall economic strength of a county/city is established. The data underscores the dynamic nature of athletic performance trends at the local government level. Through the monitoring and analysis of sporting competition models, we can validate the impact of local sports policies and training outcomes. Moreover, this study furnishes substantial data support for shaping future sports policies and making informed decisions regarding resource allocation.