This study focuses on Airborne Gamma Ray Spectrometry (AGRS) Surveying data to identify naturally existing zones with radioactive anomalies, such as potassium, uranium, and thorium, in the Wadi-Biyam and its surroundings in Egypt's Eastern Desert. The AGRS produces raw and unlabeled data collected from the background radiation and radiation emitted by the radioactive elements. Using anomaly detection statistical methods with the data produced by AGRS faces various challenges, such as inaccuracy, time consumption, bias, limited scope, and false certainty. These issues can lead to costly and hazardous outcomes, so there is a need for an innovative approach to deal with these issues. We apply two machine learning methods for anomaly detection, DBSCAN and BIRCH, with the Silhouette Coefficient to evaluate the result and mineral zone planning. Two of the nine rock types have high concentrations of radioactive elements: younger granite and wadi deposits.