Floods will always be determined and dominate by climate change. Even though there is the realization of various methods of flood control, the rate of damage is likely to grow in higher aspects unfolding the trend of the floods in the floodplain regions in future. Hence, flood control would prove valuable in case one is able to estimate the risk of floods in the future besides climate change. The riparian system of Bhagirathi River Basin in West Bengal India has been deemed as flood prone region over the last several decades. Through this paper, the flood-prone areas of the Bhagirathi basin have also been pointed out independently which can be used in disaster assessment. Susceptibility analysis is carried out using machine learning with Random Forest (RF) and Xtreme Gradient Boost (XGB) based on 12 flood inventory conditioning parameters. In the light of the tolerance of AUC, the RF model points to the sensitivity of floods in the Bhagirathi basin. Because rainfall is the principal cause of these floods the future projection of the flood risk in changing climate conditions was determined using a climate change projection in the CMIP5 based global climate models. The notion of riskiness has been examined. The resulting studies have found that several riverine blocks have been observed to be the most flood-prone risk areas in the Bhagirathi basin due to overall susceptibility in addition to high rainfall in future under the various RCPs scenarios. This study demonstrates the excellence of machine learning tool, CMIP5 and remote sensing as threat mapping potential of floods in the future.