利用社群媒體貼文特徵預測臺北富邦勇士籃球隊社群媒體參與之可行性探討

Abstract

社群媒體是現代個人與組織互動的重要平台，對於職業球隊而言，透過社群媒體參與，球迷能夠表達對球隊的認同，並加深彼此之間的關係，同時提高球隊的曝光度與球迷的忠誠度，因此，研究社群媒體參與的可行性以及分析其重要性，對於職業球隊在社群媒體平台的經營具有重要價值。本研究旨在探討利用社群媒體貼文特徵預測臺北富邦勇士籃球隊在社群媒體參與之可行性，建立多元線性迴歸 (MLR)、支援向量迴歸 (SVR)、決策迴歸樹 (DTR) 以及極限梯度上升 (XGB) 模型，利用貼文特徵 (發佈時間、歷史互動、貼文主題與類型)，預測總互動目標變量。研究結果顯示： 根據模型性能評估指標，SVR模型最適合預測臺北富邦勇士籃球隊社群媒體參與；其中，歷史互動與發佈日等貼文特徵對預測總互動之影響較大。據此，本研究建議未來可擴大研究範圍，探討不同的研究對象與社群媒體平台、納入其他特徵變量或進行模型優化，助未來社群媒體經營與發展。

Social media is an important platform for modern personal and organizational interactions. For professional sports teams, fan engagement through social media allows fans to express their support for the team and deepen the relationship between the team and its supporters. It also increases the team's visibility and fan loyalty. Therefore, researching the feasibility and analyzing the importance of social media engagement holds significant value for the management of professional sports teams on social media platforms. This study aims to explore the feasibility of predicting the engagement of the Taipei Fubon Braves basketball team on social media using post features. Multiple Linear Regression (MLR), Support Vector Regression (SVR), Decision Tree Regression (DTR), and Extreme Gradient Boosting (XGB) models were established to predict the total interaction target variable, utilizing post features such as publishing time, historical interactions, post topics, and types. The results showed that, based on model performance evaluation indicators, the SVR model was best suited for predicting the social media engagement of the Taipei Fubon Braves. Among them, historical interactions and posting date had a significant impact on predicting the total interactions. Based on this, the study recommends expanding the scope of future research to explore different research subjects and social media platforms. Additionally, incorporating other feature variables or optimizing the models could aid in the future development and management of social media marketing.