抄録
The purposes of this study were to extract the lifestyle guidelines for optimum conditions using the decision tree analysis and to show the effectiveness of the decision tree analysis on routine athletic conditioning data. The subjects were five athletes, three males and two females who were measured over a total period of 1323 days. Three factors concerning the athletes' perceived condition and nine items concerning their basic lifestyle were observed daily using a “Quality Control (QC) Sheet”. Two or more methods were tested to treat the missing value, and the method that most maintained the distribution characteristic (mean and standard deviation) of data was applied to treat missing value. Data was standardized to each subject because a significant difference was admitted in the data of each item among subjects and the data were enabled to treat on same dimension. Decision tree analysis was applied to extract the lifestyle rule for optimum conditions. Three items concerning perceived condition were dependent variables and nine items concerning basic lifestyle were independent variables. Valid classification decision trees were made, and the lifestyle guidelines for optimum condition were drawn up. For each dependent variable, the first split variable was the time subjects got up and the second split variable was the length of sleeping hours. The effectiveness of decision tree analysis for athlete's data concerning lifestyle was validated since lifestyle guidelines for optimum conditions were extracted.
