<p>The next generation of mineral exploration efforts focuses on new search spaces at greater depth in complex geological settings. At the regional scale, various mineral prospectivity analysis techniques are available to support exploration targeting, each of which has its advantages and disadvantages. Given the variety in prospectivity analysis methods, exploration targeting faces significant sources of model-related uncertainty. To address this, exploration geologists have attempted to develop and adopt robust techniques and tools for selecting targets from prospectivity models, aiming at improving the modeling performance and reducing relevant uncertainty. In this study, to generate different porphyry copper prospectivity models, eleven integration approaches, including machine learning (ML), deep learning (DL) as a subset of ML algorithms, data-driven multi-index overlay, fuzzy logic, and geometric average. Then, the confidence index (CI) ensemble approach was used to benefit from the advantages of all the models applied. Subsequently, the CI prospectivity model was converted into a binary map to identify exploration targets. The results indicate that this model identifies several high-potential areas, many of which show strong spatial correlation with known mineral occurrences. Additionally, several new targets were predicted in areas where there is no spatial correlation with known mineral occurrences. However, there are favorable geological features for the formation of porphyry copper mineralization. Consequently, according to the survival bias (SB) concept, these targets are worth further investigation, as they may indicate the presence of deeper deposits with weak signals. Overall, the application of the proposed approach has yielded reliable and reproducible results in the Varzaghan region and could be applied to other geological environments.</p>