Background/Purpose: The purpose of this study was to determine the best predictors of winning Korean pro-basketball games for men employing four statistical analysis.
Method: Korea Basketball League(KBL) informs the records of 12 variables [(2P%(rate of 2-point shots), 3P%(rate of 3-point shots), FT%(rate of free throw shots), OR(offensive rebounds), AS(assist), TO(turnovers), DR(defensive rebounds), ST(steals), GD(good defenses), BS(block shots), WFT(fouls with free throw), WOFT(fouls without free throw)], respectively. These 12 variables from a total of 2970 game records in 2001-2012 were employed to estimate winning and losing the games. Discriminant analysis, logistic regression, decision tree analysis, and artificial neural network analysis utilizing statistical software (PASW 18.0) were applied to determine the best winning estimation of the games.
Analysis/Results: The better predictors of winning the game from both discriminant analysis and logistic regression were DR, 3P%, 2P%, ST, and TO. On the other hand, the predictors from the decision tree analysis were DR, 2P%, 3P%, ST, and TO. The good estimators from the artificial neural network analysis were DR, 2P%, 3P%, TO, and ST, respectively.
Conclusions: All estimating methods showed that the five different predictors were significantly affected on the defeat of men's pro-basketball games in Korea. Although the impact of these five factors to estimate the probability of winning games were similar, the most frequently selected predictor among four statistical methods was DR.