本研究是以台灣運動彩券所提供的資料作為主要依據，來判斷美國職棒 (MLB) 在正規球季比賽的勝負，並比較加入博弈讓分後之勝負情形。隨著棒球的發展，球場上球員各項棒球數據的收集越來越詳細，常被用來判斷球隊比賽的勝負結果。
但在加入博弈讓分這個條件下，棒球數據已不足以做出較正確的預測，因此本研究收集運動彩券所提供賠率相關資料，
主要利用群集分析(Cluster Analysis)，將比賽場次分成互相沒有交集的群組。再進一步建構迴歸分類樹(Classification and Regression Trees)判別賽事結果之類別。我們採用棒球統計 Value Over Replacement Players (VORP)，收集強弱分明的比賽，透過上述流程做出較精準預測分析。

This work discusses about issues related to predictions of the results to determine win-or-lose on standard baseball season games in Major League Baseball (MLB) according to the information provided by the Taiwan Sports Lottery. Along with growth of the baseball kingdom, more and more detailed data have been collected, it is natural to use these baseball data to make predictions on the baseball game outcomes. The main objective of this study is to predict not only the original win-or-lose game results, but also results after adding the Handicap factor. Although after adding the Handicap factor, it is useful to include factor such as the up and down of the "Betting odds" before the game, in order to be able to make the real time prediction with higher accuracy in real time. Here, we use Cluster Analysis based on Value Over Replacement Player (VORP) to separate the games into three groups. Furthermore, Classification and Regression Tree methodology is used to classify game results including the Handicap factor. Through the above procedure we obtain a more accurate prediction results.

誌謝 i
摘要 ii
Abstract iii
1前言 1
2棒球 2
2.1 棒球比賽規則 2
2.2 美國職業棒球發展與現況 2
2.3 台灣運動彩券 3
2.4 棒球統計 VORP 4
2.5賠率概念簡介 4

3資料敘述 5
3.1 資料來源 5
3.2 相關名詞 5
3.2.1 強弱隊伍 5
3.2.2 賠率變數 5
3.3.3 VORP 6
4研究方法 7
4.1 迴歸分類樹 7
4.2 k-平均值集群分析 8
5研究結果 8
5.1博弈勝負與比賽勝負之區分 8
5.1.1 VORP 之迴歸分類樹 8
5.1.2 2012 賽季 VORP 迴歸分類樹分析結果 10
5.1.3 2012 賽季 VORP 迴歸分類樹分析之評估準則 11
5.1.4 2013賽季 VORP 迴歸分類樹分析結果 12
5.1.5 集群分析之迴歸分類樹 14
5.1.6 2013 賽季集群迴歸分類樹分析結果 15
5.1.7 2013 賽季集群迴歸分類樹分析之評估準則 17

6討論與結語 18
參考文獻 19
附錄一 20
附錄二 39

Breiman, L., Friedman, J.H., Olshen, R.A. and Stone, C.J. (1993). Classification and Regression Trees,Chapman and Hall, New York.

Rencher, A.C. (2002). Methods of Multivariate Analysis, 2nd. Wiley, New York.

Wu, J. (2012). Advances in K-means Clustering. Springer, Bejiing.

Baseballprospectus
http://www.baseball-reference.com

美國職棒大聯盟官方網站
http://www.mlb.com

台灣運動彩券
https://bet.i-win.com.tw