- 學位論文
Probit迴歸模型與羅吉斯迴歸模型預測理論之研究 -以美國職棒大聯盟為例
The Study on The Predictability of Probit and Logistic Regression Models: Case Study of The United States Major League Baseball
摘要
摘要 棒球在我國越來越盛行之情形下，國內大眾對於棒球有一定的認識，同時關注的不僅有國內職棒，甚至是棒球發源地的美國職棒大聯盟，加上王建民、郭泓志等人登上大聯盟，使得許多台灣球迷在賽季開打轉播時，會片刻不離的欣賞球賽。基於台灣的棒球風潮，加上球員及球賽數據透明化，本研究以美國職棒大聯盟為研究對象，其推導出為能預測賽事勝負之研究模型。研究中同時使用Probit羅吉斯模型進行估計，比較其優劣。 本研究以美國職棒大聯盟2010年例行賽之比賽數據為樣本資料，分別以Probit模型及羅吉斯模型，進行推估預測模型，並以例行賽80場比賽結果進行預測例行賽後20場之比賽勝負結果來驗證Probit迴歸模型及羅吉斯迴歸模型孰優孰劣。實證結果顯示，不管是樣本內及樣本外，在例行賽80場比賽結果預測第81~100場皆得到羅吉斯迴歸模型優於Probit迴歸模型的結果，Probit迴歸模型預測正確率較羅吉斯迴歸模型預測正確率低約5-10%，此與過去的研究顯示羅吉斯模型無論在分類能力及預設能力上皆優於Probit迴歸模型且具有一致性。
並列摘要
Abstract In recent years, baseball became more popular in Taiwan. The public in Taiwan have some knowledge of baseball, concerned with not only domestic baseball, but the birthplace of baseball - Major League Baseball. When baseball game start, many Taiwanese baseball fans will seeing it on time. Based on the baseball trend, added to the data of players and baseball game transparency. In this study, Major League Baseball as the research target, derived to the study of predict model. The comparison of advantage between Probit and logistic models. In this study, we collect the competition data of 2010 MLB regular season games as the sample data and predict the results through Probit model and logistic model to construct the forecasting model and verify which model is superior. In this paper we compare the empirical performance of the Probit regression model with the Logistic regression model. The Logistic regression model outperforms than Probit regression model both in in-sample and out-of-sample forecasts of the united states major league baseball games. Our results are consistent with that of previous studies where the Logistic regression model has better forecast abilities than the Probit regression model.