Abstract: In nature, gold typically occurs as extremely fine-grained native gold in an extremely uneven distribution. As a result, some gold deposits (ore occurrences) lack corresponding gold geochemical anomalies, posing significant challenges to geological prospecting based on geochemical anomalies. Hence, under the guidance of the geoscience big data-driven approach, this study identified the correlations among data to address geological prospecting challenges. Based on the soil geochemical data of the Gelebile River area in northeastern Heilongjiang Province, 10 elements in the area were statistically analyzed using the C_V1 and C_V1/C_V2 interpretation diagrams (C_V: coefficient of variation), revealing high gold and molybdenum mineralization potential. By constructing a regression model for gold, this study delineated two prospecting targets in the Gelebile River area. The No. 1 prospecting target was verified through trenching and drilling engineering, exposing one gold and one molybdenum ore body. This study demonstrates that the prospecting and prediction model established based on multivariate regression analysis can significantly improve prospecting efficiency. Moreover, it offers an effective solution to quantitative prediction of prospecting targets in small-scale areas where the precise locations of mineral deposits (or occurrences) remain unknown.