Riverbank erosion assessment is crucial for natural resource management and soil protection. Conventional in-situ measurements yield accurate erosion rate data; however, when applied over a vast region, the method becomes prohibitively expensive and time-consuming. This research aimed to investigate the use of an in-situ technique and machine learning (ML) to quantify bank erosion rates in alluvial rivers in Northeast India. The variables influencing bank erosion were identified and determine using Field observations and a digital elevation model analysis. The dataset was then divided into training and testing. The ML model was trained and verified using data from the Barak River in northeastern India. In-situ jet tests were employed to estimate bank erosion values on the riverbank. The ML model (GBR, DL, KNN) was employed, with the bank erosion rate as the output variable and hydraulic shear stress, bankfull discharge, bank height, soil type, vegetation index, and slope degree as the input variables. Following the training phase, the model was tested. Based on the test outcomes, the machine learning model displayed an acceptable level of capability to estimate bank erosion (R-squared: 0.95, RMSE: 2.36). The optimal model and its inputs were employed to estimate the study area's bank erosion rates. Finally, the optimal model results and ArcGIS tools created a riverbank erosion rate map.