Scene evaluation of a ball game for solving batting order optimization

Since baseball has been a big business, a scouting system becomes important for baseball teams to compose an appropriate batting order. The scouting system is required to examine a batter not only individually but also from aggregative point of view, to acquire batters who match for each lineup position. However, there are few statistical analyses into this issue so far, in spite of a long history of baseball leaves a wealth of data to use for extracting valuable regularities. Alternatively, this issue has ever relied on empirically derived know-how, which is not explicitly defined as evaluative criteria to scout an appropriate batter so far. Therefore, we propose a unifying data mining method where the required functions of each lineup position are quantitatively-extracted for giving evaluative criteria to scout appropriate batters. Our proposed method, which provides quantitative criteria of each lineup position, will be a fundamental knowledge to compose an appropriate batting order. Through some experiments, we show the effectiveness of our quantification by comparing existing methods.