<![CDATA[Selection of athletes in competitive sports is mostly based on subjective judgments; therefore, it results in inconsistency. This research presents a classification model that will help to measure the potential of athletes using the Decision Tree algorithm by utilizing real competition data from PASI DKI Jakarta. The dataset used consists of 450 records of athletes with attributes such as race category, time records, and ranking information. The analysis was performed based on the CRISP-DM framework which comprises six stages: business understanding, data exploration, preparation, modeling, evaluation, and deployment. Development and testing of the model were carried out in RapidMiner software using a 10-fold cross-validation technique. It achieved an accuracy of classification equal to 92.22% with a standard deviation of ±5.37%. The performance metrics show precision rates at 96.88% for High, 78.95% for Medium, and 94.87% for Low classes; while recall values are 100%, 88.24%, and 88.10%, respectively. The decision tree model generated specifies ranking as the root node meaning that this attribute has the highest influence on class separation among other attributes in this dataset. There are three classification rules produced by this model: ranking ≤3.500 is classified into high potential; between 3.500-6.500 belongs to medium potential; otherwise greater than 6.500 will be classified into low potential which can be applied practically as a decision support system enabling coaches to perform objective systematic data-driven processes in selecting athletes]]>
