<![CDATA[Thunderstorms pose a serious threat to aviation, including at Sultan Hasanuddin International Airport. Weather forecasters typically utilize radiosonde data, including the Showalter Index (SI), Lifted Index (LI), K Index (KI), and Precipitable Water (PW), to categorize thunderstorm occurrences. These data have varying values, which can potentially cause overlapping index distributions and subjectivity in decision-making. Therefore, the C4.5 method is needed to minimize this potential. The C4.5 method generally consists of root nodes to leaf nodes derived from gain and entropy information. This research aims to classify thunderstorm occurrences using the C4.5 method and verify them with accuracy, recall, balanced accuracy, True Skill Statistic (TSS), and Critical Success Index (CSI). The data used in this study span the period from 2013 to 2024, with a 12-hour time interval (00 UTC and 12 UTC), encompassing SI, LI, KI, and PW data sourced from radiosonde launches, as well as thunderstorm occurrence data obtained from synop codes. The data from 2013 to 2024 was then divided into two parts, namely training data (2013-2021) and testing data (2022-2024). The classification results for 2022-2024 were dominated by the non-occurrence of thunderstorms, with 1901 occurrences, while there were only 31 thunderstorm occurrences. For the overall verification results, the C4.5 method achieves a relatively good accuracy (0.785). However, recall (0.027), balanced accuracy (0.507), TSS (0.014), and CSI (0.026) have low values. ]]>
